Software engineer focused on agent infrastructure, applied AI systems.
I build practical systems around modern AI: local model runtimes, agent CI workflows, computer-use harnesses, evaluation pipelines, and backend services that are designed to be measured, tested, and deployed.
- Local and on-prem model deployment
- Agent safety, privacy filtering, and CI guardrails
- Performance benchmarking for model-backed services
- Computer-use agents and automation harnesses
- Distributed backend systems and applied machine learning workflows
| Project | Description | Stack |
|---|---|---|
| agentci-privacy-filter-harness | On-prem privacy filtering harness for agent workflows using OpenAI's privacy filter model, ONNX Runtime, and MLPerf-style benchmarking. | Python, ONNX Runtime, FastAPI, MLPerf LoadGen |
| MerchantOS | Computer-use harness for simulating agentic commerce environments for merchants. | Python |
| Agent-containers | Experiments around containerized agent execution environments. | Python |
| dev-trace | Developer tracing and observability experiments for AI-assisted workflows. | Python |
| LLMFinetuning | Fine-tuning experiments for language models across supervised and unsupervised workflows. | Jupyter Notebook, Python |
| JaiydevRAG | RAG experiments exploring retrieval patterns, vector indexes, and AI application architecture. | Python |
AI and agents: LLM workflows, token classification, RAG, fine-tuning, privacy filtering, local inference, agent evaluation
Backend systems: Python, Go, FastAPI, distributed systems, containerized services, CI workflows
Data and infrastructure: Spark, Hadoop, vector search, benchmarking, observability, deployment automation
I write about agent infrastructure, local model deployment, privacy filtering, and practical AI systems on Substack:
- Email: jaiydev799@hotmail.com
- GitHub: @gupta799