3.01
Applied Generative AI
LLM products, RAG pipelines, chat agents, and voice-driven characters — from concept to evaluated, monitored release.
ChatGPT APIRAGLangChainEleven Labs
Ghanshyam Paunikar is an AI & data practitioner based in Pune. Previously Data Analyst at Sigma AI, he designs and ships machine-learning systems across NLP, computer vision and applied generative AI — the kind that survive real traffic, not just demo day. Currently open to new opportunities.
I build AI that behaves like it's been raised properly — models that are accurate, accountable, and pleasant to be around in production. My work lives at the intersection of research, engineering, and a healthy distrust of hype.
Over the last three years I've led the full arc of ML products — from messy CSVs to 20-stream edge pipelines. I've shipped vision models with 95%+ accuracy, an SRGAN upscaler prototyped for commercial media, and ChatGPT-driven NPCs that lifted player engagement by 45%. I care about clean data, readable code, and the unglamorous work of monitoring what happens after deploy.
Now back in Pune and actively looking for my next role — ideally an AI engineering or data science position where I can own the full arc from prototype to production. Open to remote, hybrid, or on-site across India and internationally.
LLM products, RAG pipelines, chat agents, and voice-driven characters — from concept to evaluated, monitored release.
Real-time detection, tracking and super-resolution — deployed on edge devices with 95%+ accuracy across difficult conditions.
Classification, NER, embeddings and retrieval — the quiet infrastructure that makes enterprise text actually useful.
Pipelines, dashboards, A/B tests and statistical modelling — turning telemetry into decisions your team can act on.
Reproducible training, cost-sane inference and on-call-friendly observability — models that don't surprise you at 3am.
Unity, XR and conversational agents — bringing models into real-time experiences where latency and personality both matter.
7 shown · Twelve public repositories. Many more under NDA.
View all repositories ↗Work that matters, quietly. Some of the most meaningful hours I've spent weren't in a terminal.
I've volunteered with Spread Smile Foundation, contributing to initiatives focused on education, healthcare, and community development for underprivileged children. The work spanned organising fundraising events, running tutoring sessions, and helping with awareness drives — the kind of work that asks for patience more than cleverness.
I collaborated with cross-functional teams to implement sustainable solutions and make sure outreach actually reached. It taught me something engineering rarely does: that empathy and consistency compound harder than any metric, and that showing up — again and again — is its own kind of skill.
Small gestures, hands-on work, real families. Spreading smiles and a bit of hope, one weekend at a time.