Aravind E S
AI Software Engineer
Hire this person
Send a job offer directly to this candidate
About
Designed and shipped a production backend service exposing an LLM-driven agent through validated, schema-checked APIs (Pydantic), restricting it to read-only, sanitized queries to eliminate hallucinated writes to live operational data. Built a retrieval layer that fuses real-time database state with domain documentation, and added Redis-backed caching plus an asynchronous background worker to precompute reusable results without adding request latency. Owned migration of the core data layer from SQLite to PostgreSQL / Azure PostgreSQL, redesigning schemas to support concurrent, high-throughput production traffic.
Experience
AI Software Engineer - Tata Elxsi - Kerala, India
(2024-12)
- Designed and shipped a production backend service exposing an LLM-driven agent through validated, schema-checked APIs (Pydantic), restricting it to read-only, sanitized queries to eliminate hallucinated writes to live operational data.
- Built a retrieval layer that fuses real-time database state with domain documentation, and added Redis-backed caching plus an asynchronous background worker to precompute reusable results without adding request latency.
- Owned migration of the core data layer from SQLite to PostgreSQL / Azure PostgreSQL, redesigning schemas to support concurrent, high-throughput production traffic.
- Built and optimized real-time computer-vision inference services (object detection, segmentation, OCR) deployed on NVIDIA Jetson edge devices; used TensorRT/CUDA to cut inference latency by ˜25%.
- Debugged and hardened a live video-streaming pipeline (FFmpeg, GStreamer, RTSP), improving delivery stability by ˜30% under degraded network conditions.
- Shipped a safety-critical collision-detection feature that passed U.S. certification for production deployment.
AI Software Developer Intern - Tata Elxsi - Kerala, India
(2024-01 - 2024-06)
- Built radar point-cloud processing pipelines in ROS for real-time object detection.
- Extended a pretrained YOLO model with a second detection head to improve recall, without regressing existing detection accuracy.
Education
Master of Computer Applications - MCA - Cochin University of Science and Technology (2022 - 2024) Kerala, India
Bachelor of Science - Physics - Mary Matha Arts and Science College (2018 - 2021) Kerala, India