Pawan Mittal
📍 San Diego, CA
📞 1 858 288 3709
mitpawan@gmail.com
Summary
- 16 years of professional experience in software engineering with a strong focus on backend and full-stack development.
- Passionate about building data-driven systems and modeling complex, real-world behaviors.
- Experienced in designing scalable architectures for high-volume data processing and interactive visualizations.
- Strong technical writer and communicator, with a knack for mentoring, teaching, and simplifying complex ideas.
- Enjoys experimenting with code, exploring new technologies, and translating concepts into practical solutions.
Work Experience
Creator & Developer | notjustpython | Ongoing | San Diego, CA
An educational app blog demonstrating real-world software architecture using microservices and modern technologies.
- Dev Ops
- Designed and developed containerized microservices, each showcasing a different tech stack.
- Configured Nginx as a reverse proxy, with each microservice running on its own port for isolation.
- Deployed the platform on Amazon Lightsail VPS for cost-effective hosting with root-level control.
- APIs
- Built multiple backend services in Python (Flask), demonstrating RESTful APIs, JWT auth, OAuth2 login, and secure session handling.
- Implemented proof-of-concepts for core backend concepts like functional programming, data pipelines, and message queues.
- Developed demo applications and APIs using Node.js, integrated into a custom Express server.
- UI
- Built user-facing interfaces using React, styled with CSS.
- Integrated Axios for API communication across services.
- Used Docusaurus to generate static documentation-style walkthroughs with embedded code, explanations, and interactive learning.
- Managing JavaScript dependencies via npm to support evolving frontend needs.
- Mentoring
- Created hands-on tutorials that promote building, breaking, and iterating — not just reading about concepts.
- Targeted at engineers who learn best through exploration and experimentation.
Lead Python Engineer | Feature Analysis for Stock Trading (FAST) | Mar 2022 – Nov 2024 | San Diego, CA
A research-focused stock trading platform that leverages statistical models, time-series data, and machine learning to discover patterns and generate actionable trade recommendations targeting short-term gains (e.g., 5% in 1–3 days).
- Machine Learning & Feature Engineering
- Devised stock features to normalize price and volume data.
- Built a pipeline to group and scale time series data, enabling consistent model inputs across tickers and exchanges.
- Developed a CLI-driven training framework to experiment with combinations of features, lookback periods, and target definitions.
- Trained and evaluated multiple models including:
- Logistic Regression models using
scikit-learn
for binary classification of trade outcomes - Recurrent Neural Networks (RNNs) using TensorFlow, with stacked GRU layers, dropout regularization, and sigmoid output for binary classification
- Tuned RNN hyperparameters like lookback length, GRU units, dropout rate, and learning rate to avoid overfitting and improve predictive performance
- Logistic Regression models using
- Applied time series-aware feature scaling and created rolling lookbacks to model short-term directional changes.
- Defined flexible labeling logic to support variable holding periods and profit thresholds, allowing dynamic target generation.
- Evaluated model performance using metrics such as precision, recall, accuracy, and F1 score for downstream analysis and comparison across models.
- Persisted trained models to the filesystem for per-ticker evaluation and inference.
- UI
- Built a responsive, user-friendly interface using React hooks, JSX, Babel, HTML, and CSS.
- Integrated Axios for seamless API communication between frontend and backend.
- Configured Nginx to serve frontend assets and manage static/dynamic content efficiently.
- API & Data Analysis
- Engineered a robust Python API for serving trade signals and user data.
- Used Pandas for data import, transformation, analysis and statistical modeling.
- Deployed backend services using uWSGI.
- Database
- Designed schema for MongoDB collections for storing stock data, generated features, and user activity.
- Used MongoEngine ODM to abstract database access through Python classes.
- Applied design patterns including DAO, MVC, to develope a modular code base.
- DevOps
- Configured Nginx as a reverse proxy to route frontend and backend traffic across services running on different ports.