SRI RAM ESWARAN

About Me

I'm Sri Ram, computer science professional with 4+ years of experience in industrial research & development. I'm a Software Development Engineer (L5) at Amazon in Seatte, WA. I studied Computer Science at the University of Illinois Urbana-Champaign (UIUC).

I did my Bachelors' in Computer Science at SASTRA University, India and graduated with an honors. I then worked at Informatica for over three years as a Research & Development Engineer before joining UIUC for Masters.

I'm always interested in Problem Solving and you can find me thinking about strange algorithms for stranger problems. I'm a certified Advanced Problem Solver from HackerRank.

I have secured 70 citations from my 7 research articles published in conferences/seminars/peer-reviewed journals. Refer the Publications tab for more details.

I'm generally quite, love cricket and am passionate about the future of the world with the disruption coming through Aritifical Intelligence that humans develop.

I work on bringing up, maintainig, monitoring, enhancing and managing full life cycle of hosts/droplets/instances in racks across all global regions, also aimed at reducing the ratio of unsellable instances by improving observability over pipeline health.

Degree and Activities

I started my masters in Aug 2022 at University of Illinois - Urbana-Champaign.

I conduct research at Prof. Reyhan's lab on large language models (LLM). The purpose of this work is to arrive at a general fine-tuned model that can, on its own, perform code completion, debugging and test case prioritization. The study covers about 1000 open source projects, taking into account their documentations, code comments and usage for pre-training. The models primarily considered are Llama, GPT but not restricted to those, as this is an open research.

I'm also a graduate employee at Illinois Shared Administrative Service, where I develop and enhance backend portals for the U of I system. I work under the Illinois Civil Service that is responsible for the advancement of technology adaptation in various departments of the university including security, privacy of users and being green-conscious. I report to the Director of Data Services as a senior developer.

I did complete my IRB training that enables me to deal with experiments involving human subjects and behavior. For example, interviewing or studying people's behavior/response to technological changes.

I'm a member of the Tau Beta Pi Honorary Engineering Soceity that holds frequent talks, seminars, conferences among other activities.

Coursework (GPA - 4.0/4.0)

CS 597 Individual Study

  • Large Language Models (Llama, GPT) for code generation, completion, fault localization

CS 498 Cloud Computing Applications

  • Data storage, Virtualization, Cloud Analytics, Load Balancing, Hands-on tasks with cloud offerings

CS 598 ML for Software Engineering

  • Deep learning models to aid in software engineering tasks (agile), Model tasks using ChatGPT API

CS 510 Advanced Information Retrieval

  • Expectation Maximization, Topic Modelling, Conversational/Interactive IR

CS 447 Natural Language Processing

  • Dialogue Systems, Question Answering, Machine Translation, Masked Language Modelling

CS 410 Text Information Systems

  • Search Engines (Design, Ranking & Evaluation), Text Mining, PLSA, Hidden Markov Models

CS 441 Applied Machine Learning

  • Image smoothing, Convolutional Neural Networks, Recurrent Neural Networks

CS 500 Computer Education Research

  • Cognitivism, Design of Knowledge-in-Pieces, Transfer of Learning, Theoretical Constructivism

I joined Informatica as a student intern in Jan 2019. I was offered a full-time position after successful completion of my Bachelor's degree. My full-time employment started in Jul 2019 and I continued to work there till I decided to pursue my masters in Aug 2022. I spent about 3 years and 8 months at Informatica, with varying levels of duties and resposibilites throughout the journey.

My team was the Enterprise Data Catalog App Team of Informatica, which adds to Data Intelligence solutions offered by the org and almost always tops in a Gartner's Magic Quadrant category. The team broadly come under Informatica's Core Metadata Group.

I worked in the Research & Development of 10.2.2, 10.4, 10.4.1, 10.5, 10.5.1 releases of EDC along with other minor releases in between. My contributions were in front end, backend and in the CLAIRE AI engine. My primary duties included attending Product Management meetings, Development Planning, Implementation and Code Reviewing. Additional duties include resolving urgent customer issues which may be functional or deployment-related that require R & D assistance. One interesting customer-related issue was when I had to optimize an operation that blocked a signature customer from daily functions. As a result of the optimization, the running time came down from about 5-10 minutes to consistently less than two minutes. This had a huge impact in the system's ability to handling other requests. I was the spotlight award winner that quarter.

There was an internal Infa Hackathon in 2021/22, where I, as part of a 5 member team, presented an AI-powered UI (UI with which user can interact without a mouse or keyboard). Unfortunately, we couldn't win in the final round. However, it was taken into discussion and included in the code base. I was also a mentor to incoming interns in 2020/21. I led a tech debt project, where we planned to eliminate a redundant servlet layer without any change in screen behavior to the user and keeping security standards intact. We successfully completed the project and it was reviewed, later added to the product as well. All the interns secured an offer with the company. These were some projects I handled outside of regular requirements which were eventually part of our offering.

Roles (3 years, 8 months)
  • Software Engineer - Research & Development: Jul 2020 - Aug 2022 (2 years, 2 months)
  • Associate Software Engineer - Research & Development: Jul 2019 - Jun 2020 (1 year)
  • Student Intern: Jan 2019 - Jun 2019 (6 months)
Summary
  • Delivered on full-stack agile development across 5 major releases of Enterprise Data Catalog (EDC)
  • Led metadata manager that recommended intelligent column similarity decisions ( CLAIRE AI engine) for source scanners originating from structured data warehouse
  • Managed 'Application Configuration' system, worth 6 user screens for administrators in Business Intelligence (BI) during 10.5.1 release of EDC
  • Used Java, Python (and associated ML libraries), JavaScript, React.js, Node.js, MySQL, Jest Testing and Jenkins for CI/CD
Degree (Jun 2015 - May 2019)

Bachelor's Degree with Honors; Computer Science and Engineering.
Awarded the Best Outgoing Student of class 2019 from the School of Computing.

Honors Coursework
  1. Cyber Forensics
  2. Green Computing
  3. Logistics and Supply Chain Management
  4. Game Design and Development
Elective Coursework
  1. Machine Learning Techniques, Data Warehousing and Data Mining, Big Data Analytics
  2. Database Management Systems, Parallel and Distributed Systems
  3. Computer Graphics, Human Computer Interface, Discrete Mathematics (Logic)
Activities and Memberships
  1. Handled and advised a group of 20 students as a teaching staff of the 'Anukul Shiksha' Program, where seniors guide juniors with their goals in academia/industry.
  2. Assisted as a student guide in teaching and grading Data Structures, Object Oriented Programming and Design, Software Engineering, and Introduction to Programming courses.
  3. Won the 'Scalable Pattern Search and Visualization' challenge conducted by KLA Tencor by developing a scalable system that matched a polygonal pattern in a co-ordinate data space sizing in the order of TBs within 10 minutes using an approximation algorithm and parallel programming.
  4. Facilitated on-campus conferences as a student member of the Computer Society of India (CSI).
  5. Managed the release of the Cipher CSE Departmental Magazine, 2018-19 Edition.
  6. Secretary and eventually Chairman of the student's forum at SASTRA, 2018-19.
  7. Involved in teaching computer Science to underrepresented communities as a member of the National Service Scheme (NSS) of India between 2016-17.
  8. Won a gold medal on "Python Programming, Data Structures and Algorithms" from NPTEL IIT-Madras.

Google Scholar link-here

Publications

  1. Saravanan, P. et al. ‘Ensemble Gaussian Mixture Model-based Special Voice Command Cognitive Computing Intelligent System’. 1 Jan. 2020 : 8181 – 8189. [Link]
  2. Jangiti, S., E, S., Jayaraman, R. et al. Resource ratio based virtual machine placement in heterogeneous cloud data centres. Sādhanā 44, 236 (2019). https://doi.org/10.1007/s12046-019-1215-9 [Link]
  3. Jangiti, Saikishor et al. ‘Scalable Hybrid and Ensemble Heuristics for Economic Virtual Resource Allocation in Cloud and Fog Cyber-physical Systems’. 1 Jan. 2019 : 4519 – 4529. [Link]
  4. Jangiti, S., Sri Ram, E., Shankar Sriram, V.S. (2019). Aggregated Rank in First-Fit-Decreasing for Green Cloud Computing. In: Mallick, P., Balas, V., Bhoi, A., Zobaa, A. (eds) Cognitive Informatics and Soft Computing. Advances in Intelligent Systems and Computing, vol 768. Springer, Singapore. https://doi.org/10.1007/978-981-13-0617-4_53 [Link]
  5. R. Thanuja, E. S. Ram and A. Umamakeswari, "A linear time approach to detect wormhole tunnels in mobile adhoc networks using 3PAT and transmission radius (3PATw)," 2018 2nd International Conference on Inventive Systems and Control (ICISC), Coimbatore, India, 2018, pp. 837-843, doi: 10.1109/ICISC.2018.8398917. [Link]
  6. Map Reduce based Key-Word Extraction and Bagging Algorithm for Review Classification in e-Commerce Websites [Link]
  7. Eswaran, Sri Ram & Thanuja, R. & Umamakeswari, Arumugam. (2018). THREE PHASED APPROACH TOWARDS DETECTION OF BLACK HOLES IN WIRELESS SENSOR NETWORK USING TIME FACTOR (3PAT). Journal of Advanced Research in Dynamical and Control Systems. Special Issue. 197. [Link]

Software Engineering Tasks using Large Language Models (LLM)

This is a research project that is aimed towards automation of code completion, test case prioritization and debugging among other software engineering tasks. Over 1000 open source projects are to be used for fine tuning of the large language models under study. Necessary tweaks to the models are also performed. This is an ongoing project that should result in achieving the larger goal of developing a generalized model that can handle daily software engineering tasks. It may not be completely possible at this time, but the work is proceeding to clear some important milestones in this journey.

Application of Topic Mining to Browser Bookmarks

This project involved developement of an extension for Chrome browser (later to other browsers also) that enables intelligent bookmarking. It was implemented using a topic modeling function hosted on an EC2 server to automatically categorize bookmarks into meaningful divisions and ease browsing experience. Click here to view.

Analytics Service for Online Learning Platform

This was an Indian Institue of Technology - Bombay (IIT-B) Internship Project under the Ministry of Education, Government of India. I worked as part of a team that designed and implemented an 'Anaytics' section for Sunbird platform to present dashboard-based inferences to course takers and organizers. Dashboards were built separately for students, instructors and organizers. We used a number of technologies, including but not limited to D3.js, Node.js, Cassandra, HTML, CSS, Google Charts. Each member of the team had different resposibilites on a weekly rotational basis. I learned Test Driven Development (TDD) model and stakeholder communication through this project in addition to coding and designing.

Bin-packing Heuristic for Cloud Data Centers(NP-Hard Problem)

This was a project under the guidance of Dr. Saikishor Jangiti on cloud optimization heuristics. This research spanned across multiple semesters and resulted in three publications. The common idea behind these works is to present a better heuristic for Virutal Machine (VM) placement in Cloud Data Centers (CDC). This can be categorized as the classic bin packing NP-hard problem with specific constraints. The works differed in the kind of VMs to be accomodated, physical machine type and capacity, implementation, any additional real-time qualifications and so on. Savings were optimized by 22%-36% resulting in a projected cut in operational cost, thus minimizing resource wastage.
Paper 1 | Paper 2 | Paper 3

Review Classification for e-Commerce Websites

This was an academic project for the Natural Language Processing course under Prof. Pradeepa at SASTRA. The goal was to analyze sentiments expressed by users on e-Commerce websites. We used Hadoop map reduce for extraction of keywords from user reviews. The extracted words were then bagged and fed to NL tools to classify those reviews in e-Commerce websites using normal classifcation algorithmic models. We also Information theory metrices like TF, IDF, Gain, etc to confirm the correctness of the classification. This was a straightforward group project (team of three) and we presented our work in a conference in 2018, which was also published online.

Preventing DDoS Attacks in Mobile AdHoc Networks (MANETs)

The work centers around detecting and preventing black holes/worm holes in infrastructure-less networks or Mobile AdHoc Networks (MANETs). Black hole nodes in wireless sensor networks involve in packet dropping or route redirection, disturbing normal packet delivery. A collection of black holes acting in tandem create a worm hole. 3PAT algorithm handled black hole detection and an extension of 3PAT with logical assumptions resulted in a competitive algorithm for the detection of worm hole tunnels in wireless sensor networks.

Billing Software for a local retailer

This project was completed with help from my batchmate Sushmanth Reddy to aid a retail store in managing their growing demands/capacity of goods. We designed and developed a billing software using C++ and shell scripting for a retail store in Thanjavur, India. We also included an automated tax calculator/advisor functionality and customized options according to stakeholder needs. This helped the retailer (our client) expand their business much faster than how it was before. They also reported an improvement in their ability to pre-order goods, thus helping in maintaining good profit margin.

Book Trader Online

This was a website I built in 2015 that can be used for trading books. The website allows personalized logins, uploading of books, downloading of books and recommendations. These recommendations were reasonable and based on user interests (manual preferences). Wamp server was used and 5 normalized databases (Interests, Users, Books, Tracking, Trade) were maintained on MySQL to support the intended functionalities. The limitation was the presence of security issues with reset password emails and in the usage of cookies.

Data Structure Tracker

Tracker was proposed as an alternate to C++ vectors, neutralizing its cons. Initial performance, time and space complexities came out to be better than vectors/linked lists but the project did not reach its desired logical conclusion due to time constraints. Don't hesitate to contact/DM if you feel interested in this project/want to take this forward.

Current Appointments

Software Development Engineer (L5) - Amazon