Oct 2023 – Jul 2025 · 1 yr 9 mos · Gurugram, Haryana, India

• Currently building SlingShot, a GenAI-powered Engineering Copilot that transforms the Software Development Lifecycle by delivering contextual, code-aware assistance to developers. The platform combines retrieval-augmented generation (RAG), agentic workflows, and multimodal capabilities to significantly improve engineering velocity, code quality, and team productivity.
• Developed end-to-end architecture of SlingShot by combining retrieval-augmented generation (RAG) with code-dependency-aware search using Neo4j and GraphQL. Developed a multimodal RAG pipeline leveraging MongoDB for context storage, Milvus for vector embeddings, and Redis for conversation memory, deployed via FastAPI on Azure. Built an automated ingestion service with multi-repository support for GitLab and Bitbucket to ensure real-time synchronization. Integrated agentic workflows using LangChain to enable seamless interaction with Atlassian tools, supporting automated task creation and contextual SDLC assistance through custom LLMs (Claude, Gemini, OpenAI, etc.) and on-prem deployed LLMs (Llama, Pixtral, Deepseek, etc).
• Presently collaborating with the Saudi Arabian government (on-site) to pilot slingshot as part of a digital transformation initiative, leveraging Generative AI to modernize their technology landscape. With a strong emphasis on Arabic language enablement and cultural preservation.
• Slingshot is in partnership with various financial institution like Lloyds, ADCB, etc contributing to a $55M deal. for their digital transformation from legacy system.

May 2023 – Jul 2023 · 2 mos · Mumbai Metropolitan Region · Hybrid

• Partnering with growth-stage technology entrepreneurs. Investing $10-30M in each SaaS or IP-led services company.
• Current portfolio includes - SirionLabs, Blueshift, CoreStack, Mad Street Den
• Contact: sayam.singla@avatargcp.com

Jul 2022 – May 2023 · 10 mos

• Teaching assistant in the introductory course of polymer sciences at IITD

Mar 2021 – May 2023 · 2 yrs 2 mos

• Carried out the research as part of summer undergraduate research assistantship and published at Journal of Materials.
• In the latest work to design novel chalcogenide glasses, designed the machine learning models that predict the composition—property relationships of chalcogenide glasses. This covers a large spectrum of properties and composition for chalcogenides!

May 2022 – Jul 2022 · 2 mos

• Received PPO
• Developed a complimentary item recommendation system for fashion retail to complete the look based on location & trend
• Designed a model using image background for location info. and different segmentation technique for outfit feature embedding
• Applied facial Detection for person info. & Deployed the model using flask API through Docker and AWS on Bodhi platform

Mar 2022 – Jun 2022 · 3 mos

• Torch Investment is a Singapore-based Asset Management firm regulated by the Monetary Authority of Singapore (MAS). Core team has worked together for over 15 years and has executed over US$1 billion worth of transactions.
• They are a trusted advisor to eminent families and High Net Worth Individuals in Asia Pacific, Middle East, and Europe, with deep relations with leading financial institutions in these regions.
• worked on a project to
• Optimized a predictive model for identifying the top 30 performing U.S. stocks, achieving a 2.3% increase in accuracy while leveraging NAV to assess potential value growth.
• Utilized Graph Networks to analyze stock interdependencies, uncovering community insights for predicting stock performance. • Extracted tweets and used FinBERT for sentiment analysis to assess market sentiment and its impact on stock behavior.

Jun 2021 – Aug 2021 · 2 mos · England, United Kingdom

• Collaborated in a project based on "Graph Neural Networks for Recommender Systems" with Professor Hakan Ferhatosmanoglu and Ph.D. Research Scholar.
• Designed a framework guided with meta relations for information propagation in streaming GNN to get combined aspect info. for online recommendations

Jan 2021 – Mar 2021 · 2 mos · Melbourne, Victoria, Australia

• Designed the crowd density map estimator CFANet incorporated with temporal awareness and scale adaptive feature