Harshvardhan sharma

I am a practical computer vision and LLM engineer focused on delivering production-ready AI systems that scale reliably. I have driven end-to-end projects from model fine-tuning to deployment, emphasizing low latency, reproducibility, and robust monitoring.

My experience includes engineering YOLOv8 detection pipelines, fine-tuning large models (Qwen, LLaMA-family), and building retrieval-augmented generation and voice-first systems using Whisper, Ollama, and FastAPI. I automate ML lifecycles with dataset versioning, CI/CD, Docker packaging, experiment tracking, and telemetry to reduce deployment issues and improve release stability.

I collaborate cross-functionally to translate research into usable products—leading teams, securing funding, and shipping customer-facing features such as chatbots, wearable navigation, and multi-agent research assistants that improved retention, safety, and productivity metrics.