Training and Inference Acceleration

Working on Training and Inference Acceleration within the PyTorch Org.

Designing software for Meta's custom ML silicon.

Tech: C++, Python.

Perception Systems

Working on software engineering for Perception Systems.

Tech: C++, Python.

Networks Infrastructure and Delivery Systems.

Built a database schema linting service to analyze and validate data from networks and data-centers.

Designing end-to-end API using Thrift, Python, and PHP to expose networks metadata to internal UI.

Enhanced CI/CD and auditing frameworks to improve test coverage, and reduce false positive errors.

Worked on improving and building document recognition services using OCR, search patterns, and image extraction algorithms.

Also helped design and prototype a Data Lake to house unstructured data streams, and control + retrain ML models as data flows in.

Tech stack: Python, Flask, AWS platforms, inhouse + external services

Managing data driven consulting projects for DIIG's clients, aimed at improving their capability to create social impact in hand with meeting bottom lines.

Projects revolve around customer segmentation, dataset curation, web-scraping, scoring/similarity metrics, and report generation.

Worked with clients like: Ginger.io, Edstruments, & The Produce Box

Worked on the tech framework for the Paydash project, a large scale RCT to measure the impact of a mobile app in reducing payment delays and information friction in the Indian govt's MGNREGA social security scheme.

Helped build user management workflows and analyze social networks data collected from on ground surveys using fuzzy string matching.