I'm Ajay Kumar, an AI engineer and data scientist focused on Tamil language technology. I built a GPT-style Tamil name generator from scratch using Python, PyTorch, and a Docker-based, AWS-hosted stack, with a FastAPI REST API for easy integration. I'm passionate about turning language technology into practical tools for developers and researchers. In my recent work, I led end-to-end data pipelines—from scraping 40k+ Indian law Q&A pairs to building a 120k+ bare acts corpus—implementing RAG with QA, memory retrieval, and optimized inference. I shipped an interactive inference app and a robust pipeline, achieving latency under 1.5 seconds and 94-97% accuracy on key benchmarks.

Ajay Kumar

I'm Ajay Kumar, an AI engineer and data scientist focused on Tamil language technology. I built a GPT-style Tamil name generator from scratch using Python, PyTorch, and a Docker-based, AWS-hosted stack, with a FastAPI REST API for easy integration. I'm passionate about turning language technology into practical tools for developers and researchers. In my recent work, I led end-to-end data pipelines—from scraping 40k+ Indian law Q&A pairs to building a 120k+ bare acts corpus—implementing RAG with QA, memory retrieval, and optimized inference. I shipped an interactive inference app and a robust pipeline, achieving latency under 1.5 seconds and 94-97% accuracy on key benchmarks.

Available to hire

I’m Ajay Kumar, an AI engineer and data scientist focused on Tamil language technology. I built a GPT-style Tamil name generator from scratch using Python, PyTorch, and a Docker-based, AWS-hosted stack, with a FastAPI REST API for easy integration. I’m passionate about turning language technology into practical tools for developers and researchers.

In my recent work, I led end-to-end data pipelines—from scraping 40k+ Indian law Q&A pairs to building a 120k+ bare acts corpus—implementing RAG with QA, memory retrieval, and optimized inference. I shipped an interactive inference app and a robust pipeline, achieving latency under 1.5 seconds and 94-97% accuracy on key benchmarks.

See more

Experience Level

Expert
Expert
Expert
Expert
Expert
Expert
Expert
Intermediate
Intermediate
Intermediate
Intermediate
Intermediate
See more

Language

English
Fluent
Tamil
Fluent

Work Experience

AI Engineer (Tamil Name Generator Project) at Outbell Communications Private Limited
July 1, 2024 - January 31, 2026
Led enhancements including 4-bit quantization, 1M-parameter model training on 40k+ Tamil names; implemented 1st interactive inference app via REST API; integrated RAG pipeline and memory-embedding extensions; supported deployment and monitoring on AWS.
AI Engineer (Tamil Name Generator Project) at Outbell Communications Private Limited
June 1, 2020 - March 1, 2024
Led end-to-end development of a GPT-style Tamil name generator from scratch using Python and PyTorch; built data processing, tokenization, training and inference pipelines; deployed a REST API with FastAPI; implemented Dockerized AWS stack; scraped 40k+ Indian law Q&A pairs; built a 120k+ bare acts corpus; improved model capabilities with RAG and memory retrieval; achieved latency < 1.5s.

Education

B.Tech at PSNA College of Engineering and Technology
January 11, 2030 - June 29, 2026

Qualifications

Google Data Analytics Certificate
January 11, 2030 - June 29, 2026
Google Project Management Certificate
January 11, 2030 - June 29, 2026
Gemma 2B with Q LoRA + QAT
January 11, 2030 - June 29, 2026
Google Data Analytics Certification
January 11, 2030 - June 29, 2026
Google Project Management Certification
January 11, 2030 - June 29, 2026

Industry Experience

Software & Internet, Professional Services, Media & Entertainment, Education