Illia Polosukhin is a Ukrainian-born computer scientist and entrepreneur known for his work on the transformer architecture in machine learning and for co-founding the NEAR blockchain. == Early life and education == Polosukhin studied at the Kharkiv Polytechnic Institute, later relocating to San Diego and then moving to Silicon Valley. == Career == === Google and transformer research === Polosukhin worked at Google and was part of the team associated with research on self-attention that culminated in the 2017 paper Attention Is All You Need, widely credited with introducing the transformer architecture used in modern large language models. === NEAR Protocol === After his work in machine learning, Polosukhin became a co-founder of NEAR Protocol and later associated with the NEAR Foundation ecosystem. In 2023, Polosukhin publicly argued that increasingly capable A.I. systems should be more transparent and user-controlled, and expressed skepticism that conventional regulation alone would solve problems created by closed, corporate models, warning about risks such as regulatory capture. He has promoted “user-owned AI” concepts that combine open approaches with decentralized infrastructure aligned with the blockchain technology. In 2024, Polosukhin downplayed scenarios of A.I. independently causing human extinction, arguing that conflicts are driven by people and that misuse of AI would reflect human intent and incentives. Later this year, Polosukhin said the NEAR Foundation would reduce its workforce by about 40%. == Publications == Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Lukasz Kaiser, Illia Polosukhin; et al. (2017). "Attention Is All You Need". arXiv.{{cite journal}}: CS1 maint: multiple names: authors list (link)
XLeratorDB
XLeratorDB is a suite of database function libraries that enable Microsoft SQL Server to perform a wide range of additional (non-native) business intelligence and ad hoc analytics. The libraries, which are embedded and run centrally on the database, include more than 450 individual functions similar to those found in Microsoft Excel spreadsheets. The individual functions are grouped and sold as six separate libraries based on usage: finance, statistics, math, engineering, unit conversions and strings. WestClinTech, the company that developed XLeratorDB, claims it is "the first commercial function package add-in for Microsoft SQL Server." == Company history == WestClinTech (LLC), founded by software industry veterans Charles Flock and Joe Stampf in 2008, is located in Irvington, New York, United States. Flock was a co-founder of The Frustum Group, developer of the OPICS enterprise banking and trading platform, which was acquired by London-based Misys, PLC in 1996. Stampf joined Frustum in 1994 and with Flock remained active with the company after acquisition, helping to develop successive generations of OPICS now employed by over 150 leading financial institutions worldwide. Following a full year of research, development and testing, WestClinTech introduced and recorded its first commercial sale of XLeratorDB in April 2009. In September 2009, XLeratorDB became available to all Federal agencies through NASA's Strategic Enterprise-Wide Procurement (SEWP-IV) program, a government-wide acquisition contract. == Technology == XLeratorDB uses Microsoft SQL CLR(Common Language Runtime) technology. SQL CLR allows managed code to be hosted by, and run in, the Microsoft SQL Server environment. SQL CLR relies on the creation, deployment and registration of .NET Framework assemblies that are physically stored in managed code dynamic-link libraries (DLL). The assemblies may contain .NET namespaces, classes, functions, and properties. Because managed code compiles to native code prior to execution, functions using SQL CLR can achieve significant performance increases versus the equivalent functions written in T-SQL in some scenarios. XLeratorDB requires Microsoft SQL Server 2005 or SQL Server 2005 Express editions, or later (compatibility mode 90 or higher). The product installs with PERMISSION_SET=SAFE. SAFE mode, the most restrictive permission set, is accessible by all users. Code executed by an assembly with SAFE permissions cannot access external system resources such as files, the network, the internet, environment variables, or the registry. == Functions == In computer science, a function is a portion of code within a larger program which performs a specific task and is relatively independent of the remaining code. As used in database and spreadsheet applications these functions generally represent mathematical formulas widely used across a variety of fields. While this code may be user-generated, it is also embedded as a pre-written sub-routine in applications. These functions are typically identified by common nomenclature which corresponds to their underlying operations: e.g. IRR identifies the function which calculates Internal Rate of Return on a series of periodic cash flows. === Function uses === As subroutines, functions can be integrated and used in a variety of ways, and as part of larger, more complicated applications. Within large enterprise applications they may, for example, play an important role in defining business rules or risk management parameters, while remaining virtually invisible to end users. Within database management systems and spreadsheets, however, these kinds of functions also represent discrete sets of tools; they can be accessed directly and utilized on a stand-alone basis, or in more complex, user-defined configurations. In this context, functions can be used for business intelligence and ad hoc analysis of data in fields such as finance, statistics, engineering, math, etc. === Function types === XLeratorDB uses three kinds of functions to perform analytic operations: scalar, aggregate, and a hybrid form which WestClinTech calls Range Queries. Scalar functions take a single value, perform an operation and return a single value. An example of this type of function is LOG, which returns the logarithm of a number to a specified base. Aggregate functions operate on a series of values but return a single, summarizing value. An example of this type of function is AVG, which returns the average of values in a specified group. In XLeratorDB there are some functions which have characteristics of aggregate functions (operating on multiple series of values) but cannot be processed in SQL CLR using single column inputs, such as AVG does. For example, irregular internal rate of return (XIRR), a financial function, operates on a collection of cash flow values from one column, but must also apply variable period lengths from another column and an initial iterative assumption from a third, in order to return a single, summarizing value. WestClinTech documentation notes that Range Queries specify the data to be included in the result set of the function independently of the WHERE clause associated with the T-SQL statement, by incorporating a SELECT statement into the function as a string argument; the function then traps that SELECT statement, executes it internally and processes the result. Some XLeratorDB functions that employ Range Queries are: NPV, XNPV, IRR, XIRR, MIRR, MULTINOMIAL, and SERIESSUM. Within the application these functions are identified by a "_q" naming convention: e.g. NPV_q, IRR_q, etc. == Analytic functions == === SQL Server functions === Microsoft SQL Server is the #3 selling database management system (DBMS), behind Oracle and IBM. (While versions of SQL Server have been on the market since 1987, XLeratorDB is compatible with only the 2005 edition and later.) Like all major DBMS, SQL Server performs a variety of data mining operations by returning or arraying data in different views (also known as drill-down). In addition, SQL Server uses Transact-SQL (T-SQL) to execute four major classes of pre-defined functions in native mode. Functions operating on the DBMS offer several advantages over client layer applications like Excel: they utilize the most up-to-date data available; they can process far larger quantities of data; and, the data is not subject to exporting and transcription errors. SQL Server 2008 includes a total of 58 functions that perform relatively basic aggregation (12), math (23) and string manipulation (23) operations useful for analytics; it includes no native functions that perform more complex operations directly related to finance, statistics or engineering. === Excel functions === Microsoft Excel, a component of Microsoft Office suite, is one of the most widely used spreadsheet applications on the market today. In addition to its inherent utility as a stand-alone desktop application, Excel overlaps and complements the functionality of DBMS in several ways: storing and arraying data in rows and columns; performing certain basic tasks such as pivot table and aggregating values; and facilitating sharing, importing and exporting of database data. Excel's chief limitation relative to a true database is capacity; Excel 2003 is limited to some 65k rows and 256 columns; Excel 2007 extends this capacity to roughly 1million rows and 16k columns. By comparison, SQL Server is able to manage over 500k terabytes of memory. Excel offers, however, an extensive library of specialized pre-written functions which are useful for performing ad hoc analysis on database data. Excel 2007 includes over 300 of these pre-defined functions, although customized functions can also be created by users, or imported from third party developers as add-ons. Excel functions are grouped by type: === Excel business intelligence functions === Operating on the client computing layer Excel plays an important role as a business intelligence tool because it: performs a wide array of complex analytic functions not native to most DBMS software offers far greater ad hoc reporting and analytic flexibility than most enterprise software provides a medium for sharing and collaborating because of its ubiquity throughout the enterprise Microsoft reinforces this positioning with Business Intelligence documentation that positions Excel in a clearly pivotal role. === XLeratorDB vs. Excel functions === While operating within the database environment, XLeratorDB functions utilize the same naming conventions and input formats, and in most cases, return the same calculation results as Excel functions. XLeratorDB, coupled with SQL Server's native capabilities, compares to Excel's function sets as follows:
Hans Uszkoreit
Hans Uszkoreit is a German computational linguist. Hans Uszkoreit studied Linguistics and Computer Science at Technische Universität Berlin and the University of Texas at Austin. While he was studying in Austin, he also worked as a research associate in a large machine translation project at the Linguistics Research Center. After he received his Ph.D. in linguistics from the University of Texas, he worked as a computer scientist at the Artificial Intelligence Center and was affiliated with the Center for the Study of Language and Information at Stanford University. Nowadays, he is teaching as a professor of Computational Linguistics at Saarland University. Moreover, he serves as a Scientific Director at the German Research Center for Artificial Intelligence (DFKI) where he heads the DFKI Language Technology Lab. == Life and career == Hans Uszkoreit, a native of East Berlin, was actively involved in a group of young individuals who opposed the East Germany regime. His protesting against the 1968 invasion of Czechoslovakia led to his expulsion from high school and subsequent imprisonment for a period of fifteen months on charges of subversive agitation. Realizing that continuing his education in East Germany was not feasible, Uszkoreit made the decision to escape to West Berlin. There, he completed his high school education and pursued a degree in Linguistics and Computer Science at Technische Universität Berlin. During his time as a student, he worked part-time as an editor and writer for Zitty, a city magazine, which he co-founded. In 1977, Uszkoreit was granted a Fulbright Grant to further his studies at the University of Texas at Austin. During his time in Austin, he concurrently served as a research associate in a significant machine translation project. Subsequently, he received a second Fulbright grant, which enabled him to pursue a Ph.D. program in linguistics. In 1984, he successfully completed his doctoral studies, earning a Ph.D. in linguistics. Between 1982 and 1986, Uszkoreit held the position of a computer scientist at the Artificial Intelligence Center of SRI International in Menlo Park, California. In 1988, he created the Department of Computational Linguistics and Phonetics at Saarland University. In 1989 he was elected head of the Language Technology Lab at DFKI. In 2012, Uszkoreit's achievements in the domain of relation extraction led to his receipt of a Google Faculty Research Award, acknowledging the substantial progress made by Uszkoreit and his team in advancing the field. In 2013, Uszkoreit, in collaboration with Feiyu Xu and Roberto Navigli, was granted an additional Google Research Award, which provided support for a targeted project within Google's Language Understanding Program, focusing on the augmentation of language comprehension and analysis. == Personal life == He is father of a son Jakob Uszkoreit, machine learning researcher scientist, an author of the landmark paper "Attention Is All You Need", and daughter Lena Uszkoreit. == Awards == 2002 Elected Member of the European Academy of Sciences 2012 Google Faculty Research Award 2013 Google Focused Research Award
Angelo Dalli
Angelo Dalli (born 14 April 1978) is a computer scientist specialising in artificial intelligence, a serial entrepreneur, and business angel investor. == Early life and education == Dalli was born in Malta and grew up in the town of Birżebbuġa. Dalli was educated at the Archbishop's Seminary, Malta and represented Malta in the Young European Environmental Research contest held in Cologne in 1994. Dalli represented Malta in the International Olympiad in Informatics held in Eindhoven in 1995, where he won a bronze medal. Dalli started selling computer software as a teenager, and worked for the International Data Group as a freelance contributor for PC World. == Academic work == After graduating from the University of Malta, Dalli spent time lecturing on artificial intelligence and natural language processing before reading for his PhD at the University of Sheffield under the supervision of Yorick Wilks. Dalli has published over 23 peer reviewed papers in the artificial intelligence and natural language processing fields, including one of the earliest methods on timestamp extraction from documents that is now commonly used in most email applications. Angelo has also contributed to the encoding of European languages in Unicode, in particular for the Common Locale Data Repository. In the field of Bioinformatics Dalli has found a particularly useful integer sequence (sequence A062208 in the OEIS) which efficiently computes all alignments of strings of length 3 together with other generalisations (sequence A062204 in the OEIS), (sequence A062205 in the OEIS) for applications in natural language and sequence alignment. Dalli has an Erdős number of 3. Dalli has led the Maltese national informatics team in the International Olympiad in Informatics at IOI 2002 in Seoul, South Korea and IOI 2004 in Athens, Greece. == Artificial intelligence == === Trustworthy AI and Hybrid Intelligence === Angelo has been a vocal proponent of trustworthy AI that impacts society positively and believes that AI should be properly regulated. Angelo has co-founded UMNAI in 2019, with the aim of creating a new form of trustworthy AI that can explain the decisions and steps that the AI has taken to output an answer, based on a neurosymbolic AI architecture that combines neural and symbolic AI in an auditable and certain manner. === AI and society === Angelo led the Government of Malta taskforce that produced Malta's new AI regulation and national AI strategy, and is an active member of the IEEE, AAAI, ACM and the ACL. === AI in transport === Angelo had led the introduction of different machine learning techniques in intelligent transport systems (ITS), including parking, controlled vehicle access zones and dynamic traffic interchange control. His intelligent transport company, Traffiko, operated in Europe, Australia and the Middle East, and was eventually sold to Q-Free in Norway in 2015. === AI in gaming === Angelo is a well known speaker in the online gambling industry. Angelo setup one of the first companies that applied artificial intelligence in the online gambling industry, called Bit8 (now part of Intralot), with the most notable work being on algorithms that estimate and maximise player lifetime value and personalised bonusing systems. These techniques have since been widely adopted by the online gambling industry Intralot subsequently bought Bit8 in 2017. === AI and creativity === Angelo has been collaborating various artists and creatives to teach AI about creativity. The results of this collaboration is the UMA AI entity, short for Universal Machine Artist. Angelo has also co-founded the Creative Science and Arts Institute to act as a foundation for future research into AI, science, technology and creativity. UMA is creating original artwork using a modified Generative adversarial network has a third component, the human artist, to produce different learning results than standard generative AI models. The underlying discriminator in UMA started from an anti-fraud detection system and has now gradually evolved to add stable diffusion and procedural generation methods. The first two artworks generated by UMA were auctioned in October and November 2018 respectively, with all proceeds donated to charity and good causes. Ongoing work in improving UMA and furthering collaboration with other artists is ongoing. Notable exhibitions include Tomorrow's Blossoms with Selina Scerri at Esplora Museum in 2024, which explored the theme of AI and emotions. == Angel investor == Angelo is an angel investor active in the high-tech startup scene, and is a member of EBAN, and World Business Angel Forum senator. Angelo has been encouraging Maltese startups via various public events including the Zest and Budding Rockstars conferences and co-founded BAM, the Malta Business Angel network, in 2019. == Awards and honours == === Entrepreneurial and scientific === Bronze Medal, International Olympiad in Informatics (1995) Malta Top Entrepreneur Award (2019) Malta Top Entrepreneur Award (2014) WIPO IP Enterprise Award for the UMNAI Neuro-symbolic AI architecture (2022) === Corporate awards === Intralot Bit8 EGR Rising Star Award (2014) Intralot Bit8 Malta Communication Authority eBusiness Award for the Best B2B application (2015) Intralot Bit8 Malta iGaming Award for Excellence (2017)
Finite-state transducer
A finite-state transducer (FST) is a finite-state machine with two memory tapes, following the terminology for Turing machines: an input tape and an output tape. This contrasts with an ordinary finite-state automaton, which has a single tape. An FST is a type of finite-state automaton (FSA) that maps between two sets of symbols. An FST is more general than an FSA. An FSA defines a formal language by defining a set of accepted strings, while an FST defines a relation between sets of strings. An FST will read a set of strings on the input tape and generate a set of relations on the output tape. An FST can be thought of as a translator or relater between strings in a set. In morphological parsing, an example would be inputting a string of letters into the FST, the FST would then output a string of morphemes. == Overview == An automaton can be said to recognize a string if we view the content of its tape as input. In other words, the automaton computes a function that maps strings into the set {0,1}. Alternatively, we can say that an automaton generates strings, which means viewing its tape as an output tape. On this view, the automaton generates a formal language, which is a set of strings. The two views of automata are equivalent: the function that the automaton computes is precisely the indicator function of the set of strings it generates. The class of languages generated by finite automata is known as the class of regular languages. The two tapes of a transducer are typically viewed as an input tape and an output tape. On this view, a transducer is said to transduce (i.e., translate) the contents of its input tape to its output tape, by accepting a string on its input tape and generating another string on its output tape. It may do so nondeterministically and it may produce more than one output for each input string. A transducer may also produce no output for a given input string, in which case it is said to reject the input. In general, a transducer computes a relation between two formal languages. Each string-to-string finite-state transducer relates the input alphabet Σ to the output alphabet Γ. Relations R on Σ×Γ that can be implemented as finite-state transducers are called rational relations. Rational relations that are partial functions, i.e. that relate every input string from Σ to at most one Γ, are called rational functions. Finite-state transducers are often used for phonological and morphological analysis in natural language processing research and applications. Pioneers in this field include Ronald Kaplan, Lauri Karttunen, Martin Kay and Kimmo Koskenniemi. A common way of using transducers is in a so-called "cascade", where transducers for various operations are combined into a single transducer by repeated application of the composition operator (defined below). == Formal construction == Formally, a finite transducer T is a 6-tuple (Q, Σ, Γ, I, F, δ) such that: Q is a finite set, the set of states; Σ is a finite set, called the input alphabet; Γ is a finite set, called the output alphabet; I is a subset of Q, the set of initial states; F is a subset of Q, the set of final states; and δ ⊆ Q × ( Σ ∪ { ϵ } ) × ( Γ ∪ { ϵ } ) × Q {\displaystyle \delta \subseteq Q\times (\Sigma \cup \{\epsilon \})\times (\Gamma \cup \{\epsilon \})\times Q} (where ε is the empty string) is the transition relation. We can view (Q, δ) as a labeled directed graph, known as the transition graph of T: the set of vertices is Q, and ( q , a , b , r ) ∈ δ {\displaystyle (q,a,b,r)\in \delta } means that there is a labeled edge going from vertex q to vertex r. We also say that a is the input label and b the output label of that edge. NOTE: This definition of finite transducer is also called letter transducer (Roche and Schabes 1997); alternative definitions are possible, but can all be converted into transducers following this one. Define the extended transition relation δ ∗ {\displaystyle \delta ^{}} as the smallest set such that: δ ⊆ δ ∗ {\displaystyle \delta \subseteq \delta ^{}} ; ( q , ϵ , ϵ , q ) ∈ δ ∗ {\displaystyle (q,\epsilon ,\epsilon ,q)\in \delta ^{}} for all q ∈ Q {\displaystyle q\in Q} ; and whenever ( q , x , y , r ) ∈ δ ∗ {\displaystyle (q,x,y,r)\in \delta ^{}} and ( r , a , b , s ) ∈ δ {\displaystyle (r,a,b,s)\in \delta } then ( q , x a , y b , s ) ∈ δ ∗ {\displaystyle (q,xa,yb,s)\in \delta ^{}} . The extended transition relation is essentially the reflexive transitive closure of the transition graph that has been augmented to take edge labels into account. The elements of δ ∗ {\displaystyle \delta ^{}} are known as paths. The edge labels of a path are obtained by concatenating the edge labels of its constituent transitions in order. The behavior of the transducer T is the rational relation [T] defined as follows: x [ T ] y {\displaystyle x[T]y} if and only if there exists i ∈ I {\displaystyle i\in I} and f ∈ F {\displaystyle f\in F} such that ( i , x , y , f ) ∈ δ ∗ {\displaystyle (i,x,y,f)\in \delta ^{}} . This is to say that T transduces a string x ∈ Σ ∗ {\displaystyle x\in \Sigma ^{}} into a string y ∈ Γ ∗ {\displaystyle y\in \Gamma ^{}} if there exists a path from an initial state to a final state whose input label is x and whose output label is y. === Weighted automata === Finite State Transducers can be weighted, where each transition is labelled with a weight in addition to the input and output labels. A Weighted Finite State Transducer (WFST) over a set K of weights can be defined similarly to an unweighted one as an 8-tuple T=(Q, Σ, Γ, I, F, E, λ, ρ), where: Q, Σ, Γ, I, F are defined as above; E ⊆ Q × ( Σ ∪ { ϵ } ) × ( Γ ∪ { ϵ } ) × Q × K {\displaystyle E\subseteq Q\times (\Sigma \cup \{\epsilon \})\times (\Gamma \cup \{\epsilon \})\times Q\times K} (where ε is the empty string) is the finite set of transitions; λ : I → K {\displaystyle \lambda :I\rightarrow K} maps initial states to weights; ρ : F → K {\displaystyle \rho :F\rightarrow K} maps final states to weights. In order to make certain operations on WFSTs well-defined, it is convenient to require the set of weights to form a semiring. Two typical semirings used in practice are the log semiring and tropical semiring: nondeterministic automata may be regarded as having weights in the Boolean semiring. Two weighted FST can be composed. == Operations on finite-state transducers == The following operations defined on finite automata also apply to finite transducers: Union. Given transducers T and S, there exists a transducer T ∪ S {\displaystyle T\cup S} such that x [ T ∪ S ] y {\displaystyle x[T\cup S]y} if and only if x [ T ] y {\displaystyle x[T]y} or x [ S ] y {\displaystyle x[S]y} . Concatenation. Given transducers T and S, there exists a transducer T ⋅ S {\displaystyle T\cdot S} such that x [ T ⋅ S ] y {\displaystyle x[T\cdot S]y} if and only if there exist x 1 , x 2 , y 1 , y 2 {\displaystyle x_{1},x_{2},y_{1},y_{2}} with x = x 1 x 2 , y = y 1 y 2 , x 1 [ T ] y 1 {\displaystyle x=x_{1}x_{2},y=y_{1}y_{2},x_{1}[T]y_{1}} and x 2 [ S ] y 2 . {\displaystyle x_{2}[S]y_{2}.} Kleene closure. Given a transducer T, there might exist a transducer T ∗ {\displaystyle T^{}} with the following properties: and x [ T ∗ ] y {\displaystyle x[T^{}]y} does not hold unless mandated by (k1) or (k2). Composition. Given a transducer T on alphabets Σ and Γ and a transducer S on alphabets Γ and Δ, there exists a transducer T ∘ S {\displaystyle T\circ S} on Σ and Δ such that x [ T ∘ S ] z {\displaystyle x[T\circ S]z} if and only if there exists a string y ∈ Γ ∗ {\displaystyle y\in \Gamma ^{}} such that x [ T ] y {\displaystyle x[T]y} and y [ S ] z {\displaystyle y[S]z} . This operation extends to the weighted case. This definition uses the same notation used in mathematics for relation composition. However, the conventional reading for relation composition is the other way around: given two relations T and S, ( x , z ) ∈ T ∘ S {\displaystyle (x,z)\in T\circ S} when there exist some y such that ( x , y ) ∈ S {\displaystyle (x,y)\in S} and ( y , z ) ∈ T . {\displaystyle (y,z)\in T.} Projection to an automaton. There are two projection functions: π 1 {\displaystyle \pi _{1}} preserves the input tape, and π 2 {\displaystyle \pi _{2}} preserves the output tape. The first projection, π 1 {\displaystyle \pi _{1}} is defined as follows: Given a transducer T, there exists a finite automaton π 1 T {\displaystyle \pi _{1}T} such that π 1 T {\displaystyle \pi _{1}T} accepts x if and only if there exists a string y for which x [ T ] y . {\displaystyle x[T]y.} :The second projection, π 2 {\displaystyle \pi _{2}} is defined similarly. Determinization. Given a transducer T, we want to build an equivalent transducer that has a unique initial state and such that no two transitions leaving any state share the same input label. The powerset construction can be extended to transducers, or even weighted transducers, but sometimes fails to halt; indeed, some non-deterministic transducers do not admit equivalent
Amazon Elastic Compute Cloud
Amazon Elastic Compute Cloud (EC2) is a part of Amazon's cloud-computing platform, Amazon Web Services (AWS), that allows users to rent virtual computers on which to run their own computer applications. EC2 encourages scalable deployment of applications by providing a web service through which a user can boot an Amazon Machine Image (AMI) to configure a virtual machine, which Amazon calls an "instance", containing any software desired. A user can create, launch, and terminate server-instances as needed, paying by the second for active servers – hence the term "elastic". EC2 provides users with control over the geographical location of instances that allows for latency optimization and high levels of redundancy. In November 2010, Amazon switched its own retail website platform to EC2 and AWS. == History == Amazon announced a limited public beta test of EC2 on August 25, 2006, offering access on a first-come, first-served basis. Amazon added two new instance types (Large and Extra-Large) on October 16, 2007. On May 29, 2008, two more types were added, High-CPU Medium and High-CPU Extra Large. There were twelve types of instances available. Amazon added three new features on March 27, 2008: static IP addresses, availability zones, and user-selectable kernels. On August 20, 2008, Amazon added Elastic Block Store (EBS). This provides persistent storage, a feature that had been lacking since the service was introduced. Amazon EC2 went into full production when it dropped the beta label on October 23, 2008. On the same day, Amazon announced the following features: a service level agreement for EC2, Microsoft Windows in beta form on EC2, Microsoft SQL Server in beta form on EC2, plans for an AWS management console, and plans for load balancing, autoscaling, and cloud monitoring services. These features were subsequently added on May 18, 2009. Amazon EC2 was developed mostly by a team in Cape Town, South Africa led by Chris Pinkham. Pinkham provided the initial architecture guidance for EC2 and then built the team and led the development of the project along with Willem van Biljon. == Instance types == Initially, EC2 used Xen virtualization exclusively. However, on November 6, 2017, Amazon announced the new C5 family of instances that were based on a custom architecture around the KVM hypervisor, called Nitro. Each virtual machine, called an "instance", functions as a virtual private server. Amazon sizes instances based on "Elastic Compute Units". The performance of otherwise identical virtual machines may vary. On November 28, 2017, AWS announced a bare-metal instance, a departure from exclusively offering virtualized instance types. As of January 2019, the following instance types were offered: General Purpose: A1, T3, T2, M5, M5a, M4, T3a Compute Optimized: C5, C5n, C4 Memory Optimized: R5, R5a, R4, X1e, X1, High Memory, z1d Accelerated Computing: P3, P2, G3, F1 Storage Optimized: H1, I3, D2 As of April 2018, the following payment methods by instance were offered: On-demand: pay by the hour without commitment. Reserved: rent instances with one-time payment receiving discounts on the hourly charge. Spot: bid-based service: runs the jobs only if the spot price is below the bid specified by bidder. The spot price is claimed to be supply-demand based, however a 2011 study concluded that the price was generally not set to clear the market, but was dominated by an undisclosed reserve price. In 2025, AWS expanded EC2 with the compute-optimized C8gn family, powered by Graviton4 and offering up to 600 Gbit/s network bandwidth (about 30% higher compute performance than C7gn), and introduced G6f fractional-GPU instances that let customers provision one-eighth, one-quarter, or one-half of an NVIDIA L4 GPU for right-sized graphics/ML workloads. === Cost === As of April 2018, Amazon charged about $0.0058 per hour ($4.176 per month) for the smallest "Nano Instance" (t2.nano) virtual machine running Linux or Windows. Storage-optimized instances cost as much as $4.992 per hour (i3.16xlarge). "Reserved" instances can go as low as $2.50 per month for a three-year prepaid plan. The data transfer charge ranges from free to $0.12 per gigabyte, depending on the direction and monthly volume (inbound data transfer is free on all AWS services). EC2 costs can be analyzed using the Amazon Cost and Usage Report. There are many different cost categories for EC2 including: hourly Instance Charges, Data Transfer, EBS Volumes, EBS Volume Snapshots, and Nat Gateway. === Free tier === As of December 2010 Amazon offered a bundle of free resource credits to new account holders. The credits are designed to run a "micro" sized server, storage (EBS), and bandwidth for one year. Unused credits cannot be carried over from one month to the next. === Reserved instances === Reserved instances enable EC2 or RDS service users to reserve an instance for one or three years. The corresponding hourly rate charged by Amazon to operate the instance is 35 to 75% lower than the rate charged for on-demand instances. Reserved instances can be purchased with three different payment options: All Upfront, Partial Upfront and No Upfront. The different purchase options allow for different structuring of payment models, with a larger discount given to customers that pay their reservation upfront. Reserved Instances are purchased based on a resource commitment. These reservations are made based on an instance type and a count of that instance type. For example, you could reserve 100 i3.large instances for a 3-year term. In September 2016, AWS announced several enhancements to Reserved instances, introducing a new feature called scope and a new reservation type called a Convertible. In October 2017, AWS announced the allowance to subdivide the instances purchased for more flexibility. === Spot instances === Cloud providers maintain large amounts of excess capacity they have to sell or risk incurring losses. Amazon EC2 Spot instances are spare compute capacity in the AWS cloud available at up to 90% discount compared to On-Demand prices. As a trade-off, AWS offers no SLA on these instances and customers take the risk that it can be interrupted with only two minutes of notification when Amazon needs the capacity back. Researchers from the Israeli Institute of Technology found that "they (Spot instances) are typically generated at random from within a tight price interval via a dynamic hidden reserve price". Some companies, like Spotinst, are using machine learning to predict spot interruptions up to 15 minutes in advance. === Savings Plans === In November 2019, Amazon announced Savings Plans. Savings Plans are an alternative to Reserved Instances that come in two different plan types: Compute Savings Plans and EC2 Instances Savings Plans. Compute Savings Plans allow an organization to commit to EC2 and Fargate usage with the freedom to change region, family, size, availability zone, OS and tenancy inside the lifespan of the commitment. EC2 Instance Savings plans provide a larger discount than Compute Savings Plans but are less flexible meaning a user must commit to individual instance families within a region to take advantage, but with the freedom to change instances within the family in that region. AWS uses the Cost Explorer to automatically calculate recommendations for the commitments you should make how that commitment will look like as a monthly charge on your AWS bill. AWS Savings Plans are purchased based on hourly spend commitment. This hourly commitment is made using the discounted pricing of the savings plan you are purchasing. For example, you could commit to spending $5 per hour, on a Compute Savings Plan, for a 3-year term. == Features == === Operating systems === When it launched in August 2006, the EC2 service offered Linux and later Sun Microsystems' OpenSolaris and Solaris Express Community Edition. In October 2008, EC2 added the Windows Server 2003 and Windows Server 2008 operating systems to the list of available operating systems. In March 2011, NetBSD AMIs became available. In November 2012, Windows Server 2012 support was added. Since 2006, Colin Percival, a FreeBSD developer and Security Officer, solicited Amazon to add FreeBSD. In November 2012, Amazon officially supported running FreeBSD in EC2. The FreeBSD/EC2 platform is maintained by Percival who also developed the secure deduplicating Amazon S3-cloud based backup service Tarsnap. Amazon has their own Linux distribution based on Fedora and Red Hat Enterprise Linux as a low cost offering known as the Amazon Linux AMI. Version 2013.03 included: Linux kernel, Java OpenJDK Runtime Environment and GNU Compiler Collection. On November 30, 2020, Amazon announced that it would be adding macOS to the EC2 service. Initial support was announced for macOS Mojave and macOS Catalina running on Mac Mini. === Managed Container and Kubernetes Services === Amazon Elastic Container Registry (ECR) is a Docker registry service for Amazon EC2
AI Pair Programmers Reviews: What Actually Works in 2026
Curious about the best AI pair programmer? An AI pair programmer is software that uses machine learning to help you get more done — it combines speed, accuracy, and an interface that just works. Hands-on testing shows real-world results vary, so a short free trial is the smartest way to decide. Whether you are a beginner or a pro, the right AI pair programmer slots into your workflow and pays for itself fast. Read on for hands-on impressions, pricing tiers, and the standout features that matter.