Jun 2023 – Aug 2024 · 1 yr 2 mos · Remote

• Developed a mentor-mentee matching recommender system at a
• startup, leveraging client-side language models. LOR from my manager at Microsoft https://drive.google.com/file/d/1ycZotmIe2Zm1l6fuhtMRqYn3ivrUaft8/view

Feb 2021 – May 2023 · 2 yrs 3 mos · Bengaluru, Karnataka, India · On-site

• Compressed BERT/GPT2 styled models for efficient edge-device inference with
• techniques such as pruning and quantization resulting in 10x latency improvement (1s to 100ms) and 2x less memory requirement (~200MB to 10MB). Benchmarked with optimization frameworks such as NVIDIA's TensorRT.
• Developed a system handling planet-scale traffic and has been optimized for GPU memory, GPU throughput, and network latency for image super-resolution called DeepEnhance within the Edge browser, using CNNs/Vision transformer architecture.
• I developed services capable of handling planet-scale traffic to generate user
• suggestions in the Bing chatbot powered by GPT-4. Designed and maintained the
• suggestion recommendation system for Bing-ChatGPT integration, and
• managed production live sites.

Jan 2020 – Jan 2021 · 1 yr · Bengaluru, Karnataka, India · Hybrid

• Built a risk management platform for risk management of structured
• mortgage products and stress testing with different stress scenarios.
• Built and benchmarked CVA(Credit Valuation Adjustment) computation using Deep Neural Networks and American Monte Carlo.

May 2019 – Jul 2019 · 2 mos · Bengaluru, Karnataka, India · On-site

• Credit valuation adjustment with neural networks and monte-carlo. Design and stress-test pricing systems.