Work go back

Engineer III

American Express (via IntraEdge)

February 2025 till present • Bengaluru, IN

  • Built CC Listener module for Bridge Intelligence, capturing real-time CCs captions for thousands of Webex meetings, streaming them into Redis for downstream summarization using GPT-4o.
  • Deployed CC Listener to Amex’s Hydra platform (RedHat OpenShift) across dev, QA, and pre-prod using Helm, Jenkins, and Docker; navigated complex internal infra and CI/CD pipelines.
  • Built a hybrid semantic + fuzzy (trigram) retrieval system over 2M+ Confluence pages (runbooks); reduced latency from 30s (native search) to under 2s.
  • Designed a multimodal runbook generation pipeline using GPT-4o, converting raw incident data (text + screenshots) into standardized Confluence pages.
  • Tech stack: _Python, FastAPI, Langchain, PGVector, Streamlit, Node.js, Puppeteer, WebEx JS SDK, OpenShift_

Associate Software Engineer

Fiery, formerly known as EFI (an Epson company)

Jul 2023 till February 2025 • Bengaluru, IN

  • Fine-tuned Mistral-7B using QLoRA on 4070 Ti Super to incorporate internal documentation and company workflows.
  • Deployed the fine-tuned model via vLLM on an NVIDIA T4 cluster; leveraged dynamic batching and quantized inference for sub-second 95th-percentile latency (TTFT).
  • Led development of 'Beacon', a print-request automation tool powered by a fine-tuned ModernBERT NER model; parsed email and chat inputs to trigger document workflows.
  • Built AskDB, an AI agent used by 300+ internal users; used llama 3.1 to translate business queries into SQL, fetch data and auto-generate charts and summaries.
  • Tech stack: _vLLM, Python, PyTorch, QLoRA, peft, SFT, Mistral, Llama, SQL, FastAPI_

Data Scientist, Intern

Fiery, formerly known as EFI (an Epson company)

Jan 2023 till Jul 2023 • Bengaluru, IN

  • Created a product mockup pipeline using ImageMagick and Node.js, achieving cost-effective design automation.
  • Tech stack: _Node.js, ImageMagick, Angular_

Fullstack Developer, Intern

Corteva Agriscience

Jul 2022 till Dec 2022 • Hyderabad, IN

  • Migrated a Flask monolith to containerized microservices (Docker, Kubernetes), reducing AutoML job-launch latency by 40%.
  • Tech stack: _Python, Flask, Docker, Kubernetes._