// Hello, I'm DeepKickoff — Hello, I'm Deep

Software Engineer
Specializing in ML & Web

Development lead who owns delivery end-to-end — turning ambiguous, customer-facing problems into shipped software, from discovery to production.
Now building LLM agents.

// skills.jsonThe SquadTechnical Skillset

Languages

JavaScriptPythonHTML/CSS/SCSSSQLC#.NET Framework

AI & Machine Learning

TensorFlowPyTorchOllamaRAGNLPComputer Vision

Frameworks & Libraries

ReactNext.jsPyTorchLangchainBeautifulSoup

Data & Tools

REST APIsGoogle Maps APIBigQueryLookerWeb ScrapingData Modeling

Systems & Cloud

GitDockerAWSLinuxCI/CD

// experience.mdMatch ReportProfessional Journey

Software Engineer

Sun Net Inc.July 2021 - Present

Architected a central entity linking system decoupled from 5 legacy modules, increasing platform modularity. Designed workflows saving customers 80%+ time. Led end-to-end development of REST APIs and javascript frontends. Mentored junior engineers. Building agentic applications in the utility space.

Application Developer Intern

Commonwealth Care AllianceMay 2019 - May 2021

Designed Provider Directory backend using .NET and Google Maps API. Built data pipelines to load cybersecurity data into BigQuery. Created executive Looker dashboards for security visualization.

// education.mdTraining GroundAcademic Background

Master of Science in Artificial Intelligence

San Jose State University2024 - 2026

Relevant coursework: Recommender Systems, Data Mining, Deep Learning, NLP, Reinforcement Learning, Artificial Intelligence and Data Engineering

Bachelor of Science in Computer Science

California State Polytechnic University, Pomona2017 - 2021

Minor in Entrepreneurship. Dean's List. Relevant coursework: Data Structures and Algorithms, Operating Systems, Database Systems, Web Development and Software Engineering.

// projects/HighlightsFeatured Projects

HealthVA — Multi-Agent Health Assistant

PythonLangGraphPostgres RLSGuardrail EvalsLocal LLM

A health assistant that routes each message to the right specialist — diet, workout, or general — so the model only sees the tools and instructions it needs for that turn. User profiles persist across sessions.

Safety is built into the system, not left to the model: answers must be backed by real tool calls or they get retried, every decision is logged, and an injury guardrail passed a published evaluation with zero unsafe exercises across 5,757 prescribed.

User data stays protected — no personal identifiers are stored, and each session can only touch its own user's data, enforced both in code and in the database. Runs on cloud models or fully local ones via Ollama.

Provider Directory

.NET CoreGoogle Maps APISQL

Designed and built the backend for Commonwealth Care Alliance's provider search tool.
Integrated Google Maps API for geolocation-based search and optimized SQL queries to handle thousands of provider records with low latency.

Diet Planning Agent

PythonLangGraphSQLiteLocal LLM

A conversational diet-planning agent running completely locally using LangChain + LangGraph. Features include BMR/TDEE calculation, food lookup via local FoodData Central, and recipe search over a local SQLite database.

Designed with safety guardrails and tool-calling capabilities to provide grounded nutritional advice without external dependencies. Its diet capabilities were later folded into HealthVA above.