Deepak Tatyaji Ahire

Gen AI Tech Lead @Bajaj Finserv | Ex - Microsoft, Roche, HelpShift

Pune, IN · Website · LinkedIn · GitHub · kaggle.com · researchgate.net · stackoverflow.com

Open to consulting, cofounder, jobs, speaking

Machine Learning Engineer with over 7 years of hands-on experience in machine learning, software engineering, and generative AI. Expertise in end-to-end ML pipelines, distributed training, production inference, MLOps, LLM fine-tuning, RAG, and real-time conversational AI. Kaggle Notebooks Expert and CodeChef CCDSAP Advanced certified. All India Rank 1 in NPTEL’s Essential Mathematics for Machine Learning (2022) and Introduction to Machine Learning (2023). B.Tech gold medalist with published algorithms research. English: C2 Proficient.

Experience

• 2025-10 to now
  Lead Software Engineer II (GenAI), Bajaj Finserv Direct Limited

  Designed an enterprise AI voicebot integrating STT, LLMs, TTS, telephony, and live-agent transfers. Built C++ streaming pipelines, Django services, and analytics; optimized latency and scalability.

• 2025-06 to 2025-10
  Staff Machine Learning Engineer, Sema — via Deel/HireGlobal

  Architected Graph RAG document retrieval and production LangChain applications with schema validation, rate limiting, and token-aware batching. Full-time via Deel (HireGlobal), Employer of Record.

• 2024-07 to 2025-05
  Senior Machine Learning Operations Engineer, Roche

  Deployed Neptune.ai on AWS EKS and migrated on-premises data. Automated infrastructure with Terraform, Argo CD, and Helm. Led an award-winning Finance Co-pilot at the Roche x Microsoft AI Hackathon.

• 2024-01 to 2024-06
  Senior Machine Learning Engineer, Sema — via Velotio Technologies

  Developed code chunking for 30 languages, knowledge-graph RAG, and PDF parsing. Implemented PEFT-based LLM pre-training and fine-tuning with Hugging Face Transformers. Via Velotio Technologies.

• 2022-01 to 2023-09
  Machine Learning Engineer, Helpshift

  Developed intent-recognition models and distributed ML pipelines. Migrated models to MLflow without downtime; optimized model management to save over 5,000 AWS EC2 compute minutes per month.

• 2020-01 to 2021-05
  Collaborative Researcher, Walchand College of Engineering

  Published research on Kth maximum-sum pairs and shortest-path algorithms in directed acyclic graphs. Developed parallel graph-processing implementations using C++ and OpenMP.

• 2018-06 to 2019-09
  Software Engineer, Microsoft India R&D

  Developed Device Care microservices, telemetry, test automation, and service-order workflows using .NET, C#, Power BI, and Azure. Built reusable Azure DevOps components for code quality and coverage.

Education

• 2018
  Walchand College of Engineering, Sangli, B.Tech; Gold Medalist (2018), Computer Science and Engineering
• 2020
  University of California, San Diego, M.S. program (Dropped Out), Machine Learning and Data Science

Skills

• python
• cpp
• machine-learning
• generative-ai
• voice-ai
• rag
• knowledge-graphs
• mlops
• llm-fine-tuning
• pytorch
• tensorflow
• scikit-learn
• hugging-face-transformers
• langchain
• llamaindex
• mlflow
• docker
• kubernetes
• aws
• azure
• terraform
• postgresql
• neo4j
• qdrant

Languages

• en
• hi
• mr

Links

• Website: https://adeepak7.github.io/
• LinkedIn: https://www.linkedin.com/in/adeepak7/
• GitHub: https://github.com/adeepak7 unverified
• kaggle.com: https://www.kaggle.com/adeepak7
• researchgate.net: https://www.researchgate.net/profile/Deepak-Ahire-3
• stackoverflow.com: https://stackoverflow.com/users/7422352/deepak-tatyaji-ahire