Usability & Target Audience First
I start by defining the user, the decision they’re trying to make, and the moments where confusion shows up. The best engineering still fails if the interface doesn’t respect attention and context.
M.S. Computer Science at UC Davis. Focused on machine learning, data systems, and security conscious engineering.
Hi, I’m Hima.
I’m warm, curious, and quietly ambitious — the kind of builder who cares about the last 10%: how something feels to use, how it behaves under real data, and how it holds up under security review. Also: I’m a cat person.
I build thoughtful, secure systems that turn complex data into usable decisions.
How I Build
I like simple surfaces with strong foundations. I prefer clean UX, careful data handling, and a security posture that’s intentional, not performative.
Simple but Powerful UI
I design for clarity: strong typography, honest layouts, and just enough interaction to make complexity feel approachable.
Secure by Default
I build with OWASP-style thinking: safe inputs, least privilege, and predictable failure modes. I’m also CWE-aware, so I treat security as an engineering constraint from day one.
Creative & Data-Driven
I enjoy turning messy signals into understandable stories with reliable pipelines, thoughtful evaluation, and visualizations that earn trust through transparency.
Focus Areas
Four lanes I keep returning to because they’re where product impact and real engineering meet.
Machine Learning & Data Systems
What: Neural Networks, CNN,Supervised Learning, Unsupervised Learning, Time-series modeling, evaluation design, and reliable feature/data pipelines.
Why I care: The best models are the ones you can trust, reproduce, and ship.
Coursework:
- ECS 271 Machine Learning & Discovery
- Artificial Intelligence
- Linear Algebra
- Engineering Mathematics
- Stanford CS229, MIT 6.S191 (YouTube deep dives for ML)
Software Engineering
What: Clean interfaces, services, dashboards, and developer-friendly systems.
Why I care: I love the craft: sharp edges removed, friction reduced, reliability increased.
Coursework:
- ECS 260 Software Engineering
- Operating Systems
- Data Structures and Algorithms
- Distributed Computing
Security & Post‑Quantum Cryptography
What: Security assumptions, attack surfaces, and performance trade-offs — including PQC.
Why I care: Security is user trust. It’s not optional — it’s the baseline.
My Bachelor's Thesis: Explored how cryptographic systems must evolve in a post-quantum world. I examined NIST-selected algorithms such as Falcon, Kyber, Dilithium, and SPHINCS+, analyzing their security models, computational trade-offs, and implementation challenges. The project deepened my interest in security as both mathematics and engineering discipline
Coursework:
- Cryptography and System Security
- Ethical Hacking and Security
- Cybersecurity and Laws
- Computer Networks
Analytics & Decision Systems
What: SQL analytics, dashboards, and turning raw activity into real decisions.
Why I care: I’m happiest when data becomes a clear next step.
Coursework:
- Applied Data Science
- STA 220 Data & Web Technologies
- Big Data Analytics
- Datawarehousing and Mining
Experience
Student Developer Assistant — Development and Alumni Relations, UC Davis
- Designing and maintaining internal analytical systems for operational and engagement insights
- Building dimensional data models (Snowflake schema) across SharePoint and Microsoft Dataverse with reliable extraction and validation
- Developing secure, role based dashboards for cross functional teams to monitor engagement, operations, and RFP tracking
Research Engineer Intern — JobRobo
- Built Python based pipelines to curate and standardize 1,000+ startup records for downstream modeling
- Applied clustering, association rule mining, and statistical methods to improve analytical accuracy by ~25%
- Designed a quantitative TAM estimation framework, reducing market research turnaround time by ~30%
Selected Work
These are some projects which I have built during my journey as a developer. My motivation to do a project has always stemmed from a problem that I have recognized around me.
Financial Markets Analysis
Problem: Most “finance ML” demos look impressive but quietly break time-series rules and produce misleading results. I wanted to allow people from all backgrounds to explore Financial Stock Market Datasets and visualize the data in a way that is easy to understand and use.
Solution: Built a local-first workflow with dataset validation, time-series-safe evaluation, and a clean API surface (dashboard + backend) to keep experiments reproducible.
Impact: A trustworthy baseline for iterating on models without losing rigor.
Spotify Automation
Problem: My friends always complained that they loved tracks of Spotify Discover Weekly Playlist but had to save them all the time. Great recommendations disappear fast and “liking” tracks isn’t the same as building an intentional library.
Solution: Built an OAuth-based app that safely saves Discover Weekly tracks into a curated playlist with clean token handling.
Impact: Turns weekly discovery into a durable, searchable archive.
Interactive Learning Platform (PGM Virtual Lab)
Problem: There was no online interactive lab for Probabilistic Graphical Models and my professor had to use different platforms. The students learned theory but lacked a visual, hands-on environment for Bayesian networks, Markov chains, or inference.
Solution: Built a full-stack platform (React + lightweight REST backend) for real-time visual modeling and execution, with interactive Python modules for Bayesian modeling, Markov chains, and MLE. Later extended it into a cross-platform mobile app using GoNative.
Impact: Enabled hands-on probabilistic learning and expanded access beyond the classroom.
LLMs for CWE (Ongoing Research) Ongoing
Problem: When LLMs generate code, what types of security vulnerabilities emerge and how do they map to established CWE categories? Additionally, how do different prompting strategies influence whether generated code is secure, insecure, or misleading?
Solution: Systematically analyzing LLM generated code against CWE standards, evaluating vulnerability patterns, and experimenting with structured prompting techniques to measure how constraints, context, and security guidance affect output quality and risk exposure.
Impact: Developing a structured evaluation approach for understanding how prompt design influences software security and identifying patterns that lead to safer AI generated code.
Building with AI, Intentionally
I use AI like a power tool: with structure, intent, and a clear definition of “done.”
Cursor as a workflow, not a shortcut
I don’t “vibe code.” I treat AI as a productivity layer, not a substitute for thinking. I design the system architecture, write core functionality and API logic myself, and use AI to accelerate repetitive or mechanical tasks such as styling refinements, refactoring patterns, and documentation.
Evaluation first
I approach AI tools the same way I approach models: define the objective, test assumptions, measure output quality, and refine. Efficiency is valuable, but only when results are reliable.
Continuous learning and upskilling
I stay current by reading research papers, exploring new tooling, and studying early stage startups including companies emerging from Y Combinator. I’m interested in how ideas move from research to real products and I actively experiment with new tools to understand their practical limits.
Built on real experience
Three years ago, I was already analyzing AI startups and building ML-driven market intelligence models as a Research Engineer Intern so my relationship with AI is long-term, practical, and grounded in outcomes.
Connect
I am open to new opportunities and always enjoy thoughtful conversations with founders building user centered products especially in data driven systems and finance.