👨‍💻 About Me I'm a Software Engineer working on AI infrastructure, currently an MS in Data Science student at CU Boulder (GPA 3.99/4.0). Two years of production engineering at Oracle. Now I work on the systems layer underneath LLM applications: how models get served, how agents get evaluated, and how GPUs get shared. 🔭 Building multi-LoRA serving, agent evaluation harnesses, and KV-cache-aware inference on vLLM 🛠️ Interested in inference performance, GPU resource sharing, and serving-stack reliability 🏥 Ex-Stanford Health Care (AI Developer Intern) · 🏦 Ex-Oracle (2 yrs, core banking at scale) 🏆 Red Bull Innovation Track Winner, HackCU12 📬 Open to AI infrastructure / ML platform roles from May 2027

Multi-LoRA Serving Platform (in progress) Serving many LoRA adapters from one shared base model on pooled GPUs, with an adapter-aware request router that sends each request to an instance already holding the adapter it needs, cutting adapter load overhead. vLLM for serving, Kafka for job queueing, Redis for adapter and job state, Spark for dataset preprocessing, Kubernetes for GPU fine-tuning jobs, GCS for adapter storage.

Agent Harness Evaluation (in progress) Running an open-source coding agent on Harbor evaluation tasks against a self-hosted vLLM endpoint, measuring how harness design (full conversation history vs. summarized context) trades serving cost against task success.

KV-Cache-Aware LLM Serving (in progress) Benchmarking vLLM under multi-turn agent traffic on a MIG-partitioned A100, instrumenting TTFT, p99 latency, cache occupancy and queue depth to find where throughput and tail latency break down.

Serving & GPU: vLLM · LoRA/PEFT · PagedAttention · CUDA · NVIDIA A100/MIG · Slurm · Apptainer · PyTorch · Prometheus Agents & LLMs: LangGraph · LangChain · Tool-Calling · ReAct · Pydantic · Multi-Agent Systems · Claude Code · Devin Data & Platform: Kafka · Spark · Airflow · Redis · Kubernetes · Docker · PostgreSQL · FastAPI · GCP · AWS · Azure Languages: Python · C/C++ · SQL · PL/SQL · Scala · Shell/UNIX

🏥 AI Developer Intern, Stanford Health Care (May 2026 – Aug 2026)

• Built a real-time BACnet/IP telemetry pipeline and database from scratch for hospital infrastructure, feeding ML and agentic workflows
• Deployed on-premise LLM agents with tool-calling and hardened error handling, and owned serving-stack uptime and health monitoring in a regulated environment

🏦 Software Developer, Oracle Financial Services (Aug 2023 – Aug 2025)

• Owned core banking modules serving 600+ institutions (UBS, Access Bank), building data pipelines in Python, PL/SQL, Shell and REST
• Automated an ETL and validation framework over 10K+ daily transactions, cutting defects 30%, and ran containerized deployments at 99.9% uptime

📊 Data Engineer Intern, The Thick Shake Factory (May 2022 – Jul 2022)

• Built Kafka streaming and Airflow ingestion pipelines consolidating 40+ outlets into columnar storage, powering real-time dashboards over 100K+ daily transactions