Software engineer at Cisco building AI platform systems across model serving, retrieval-grounded agents, LLM evaluation and replay, and cloud-native infrastructure.
Master of Computer Science student at UIUC · Open to: Applied AI / AI Platform · Backend · Infrastructure · Forward-Deployed Engineering
- Model infrastructure: GPU-backed vLLM serving, Qwen3 embeddings, and LiteLLM/OpenAI-compatible model integrations.
- Grounded agent systems: RAG and MCP retrieval with diagnostic context; developing direct source-code inspection and evidence guardrails.
- Evaluation and reliability: defect- and code-derived ground truth, deterministic and LLM-as-a-judge scoring, Temporal replay, PostgreSQL/S3 persistence, Grafana observability, and CI-integrated regression testing.
OPEN SOURCE · JAVA/SWING
Built a dialog and runner for configuring, executing, and stopping Couchbase Pillowfight workloads from JetBrains IDEs.
View merged PR #20 →
TEAM CONTRIBUTION · REACT/TYPESCRIPT
Implemented route-aware discovery and detour filtering with Google Maps/Places, plus Firebase persistence for routes and Gemini-assisted chat.
Explore five merged PRs →
APPLIED AI · DJANGO
Built pantry-aware recipe generation, authenticated recipe storage, and recipe-scoped cooking assistance.
Explore the project →
More experiments: QuantumGameHub — four simulator-backed Qiskit demos exploring interference, entanglement, Bernstein–Vazirani, and measurement-driven art.
- Models and inference: vLLM · Qwen3 embeddings · LiteLLM · AWS Bedrock · OpenAI-compatible APIs
- Agents, retrieval, and evaluation: Python · MCP · RAG · LanceDB · LLM-as-a-judge
- Backend and orchestration: Java · TypeScript · Temporal · PostgreSQL · S3
- Platform and operations: Kubernetes · Helm · Docker · AWS · Grafana · GitHub Actions

