Experience & Education

Building AI agents for chip design verification flows

Building AI-powered solutions from the ground up

Key achievements

• Built core integrations and infrastructure for MCP, Voice Assistants, and document understanding flows
• Built eval harnesses for evaluating and improving the performance of the agents platform
• Built core tracing / observability infrastructure for the agents platform

Building data pipelines + models for building web agents

Key achievements

• Implemented data pipelines for collecting multimodal text-image data for SFT and DPO datasets for VLMs and scalable recommendation systems utilizing vector, graph, and relational databases
• Implemented API endpoints, observability, and general business logic on our Rust backend. Used Python for data collection / augmentation, and adding custom logic to our Python SDK, and other customer specific integrations

Working on formal verification of software and hardware systems

Key achievements

• Worked on the Koika project writing code for verifying security properties on multiprocessor machines using Coq. Working on the ATL project formally verifying tensor compilers on C programs in Fall 2024.

Working on embedded firmware for baremetal servers

Key achievements

• Worked for AWS Hardware Engineering Services in the BIOS+UEFI Firmware Development team working on optimizing hardware interrupts from CPU-BMC communication over SSIF interface on Intel-based baremetal servers.
• Worked with Linux kernel to perform telemetry on hardware interrupts and automate firmware testing, and built API servers / microservices with Docker for data access. Used C, Python, x86 Assembly with embedded Linux, AWS CDK and SQL. Improved data infrastructure and optimized hardware interrupt processing by a 15 percent margin.

Working on a web application for tracking cryptocurrency transactions

Key achievements

• Developed gradient boosting models, hyperparameter tuning methods and bagging classifiers in model construction. Used Apache Spark and Parquet for data engineering and developing ETL pipelines to aid with inference and building training datasets. Used Go-Ethereum for fetching blockchain data and other Web3 technologies.