Jaini Shah

Design. Build. Deploy. Scale.

Jaini ShahFull-Stack Software Engineer · Los Angeles, CA

I'm a full-stack software engineer with 4+ years of designing, building and deploying production applications across fintech, consumer AI, logistics and enterprise HR. I own features end to end, from requirement gathering and system design to deployment and monitoring.

On the frontend, I build responsive, data-rich interfaces for web and mobile. On the backend, I design RESTful APIs and microservices with clear contracts, authentication, validation and documentation, backed by well-modeled relational and NoSQL databases. Where it adds value, I integrate AI: LLM-powered classification, RAG and computer-vision pipelines, with clean data going in and verified output coming out.

I deploy with containers, infrastructure as code and automated CI/CD on AWS and Azure, and I design for scale and failure with caching, event streaming, testing and observability built in. Along the way I've led Agile teams and mentored developers.

• Requirements
• →
• System design
• →
• UI
• →
• RESTful APIs
• →
• Database
• →
• LLM / RAG
• →
• Tests & CI/CD
• →
• Deploy
• →
• Monitor

The way I build

The stack changes from project to project. The engineering discipline does not.

Where I've shipped

Four-plus years across fintech, consumer AI and enterprise software, owning features from specification to production.

1. May 2025 – PresentCurrent

   Full Stack Developer

   Plaid Inc.

   Engineering the AI-Powered Plaid Insights Portal, which gives fintech clients categorized transactions, predictive financial insights and real-time analytics on top of their connected financial data, so client teams can automate financial workflows instead of reviewing and categorizing transactions by hand.

   • +12%classification accuracy
   • −18%API error responses
   • 28%faster dashboard loads

   • Claude
   • Python
   • FastAPI
   • Celery
   • Pandas
   • Pydantic
   • React
   • PostgreSQL
   • AWS
   • Terraform

   Full scope of work

   • Product definition: ran requirement-gathering sessions with stakeholders and translated fintech client needs into technical specifications, feature priorities, user flows and integration points with backend microservices.
   • Data preparation: built the pre-processing stage for raw transaction data: normalizing merchant descriptors, amounts, currencies and dates, removing duplicates and pending/posted overlaps, and validating every record against Pydantic schemas before inference.
   • Model integration: integrated Claude for transaction categorization and predictive insights with the AI engineering team, exposing model endpoints through Python, FastAPI and REST, improving classification accuracy by 12%.
   • Output verification: added post-processing that validates model responses against a strict JSON schema, applies confidence thresholds and rule-based fallbacks, and logs low-confidence cases for review before results are persisted.
   • Microservices: developed account-aggregation and payment-processing services with FastAPI and Celery, structured into presentation, service and data layers and deployed on AWS Lambda and S3.
   • Secure APIs: designed RESTful APIs with OAuth2, JWT, rate limiting, encryption and structured logging to meet regulatory standards, reducing error response rates by 18% and integrating reliably with Plaid's financial endpoints.
   • Analytics dashboards: built React dashboards on AI-enhanced APIs and PostgreSQL (AWS RDS) with optimized state management, caching and lazy loading, improving load times by 28%.
   • Quality and delivery: unit, integration and validation testing with PyTest and Jest; automated CI/CD with GitHub Actions and Terraform, deploying to AWS EC2 with DevOps for reliability and observability.
2. Oct 2024 – Apr 2025

   Software Developer

   MealMate

   Built an AI-powered meal-analysis app that turns a food photo, barcode or plain-text description into calories, macronutrients and ingredients in real time for more than 1,000 users.

   • 1,000+active users
   • −40%p95 latency
   • +30%successful analyses

   • YOLOv8
   • EfficientNet
   • spaCy
   • OpenAI
   • ONNX Runtime
   • Python
   • React Native
   • Redis

   Full scope of work

   • Computer vision pipeline: YOLOv8 detects and localizes each food item in a photo; an EfficientNet-B0 classifier fine-tuned on Food-101 identifies the dish; portion size is estimated from the item's area relative to the plate.
   • NLP pipeline: spaCy named-entity recognition and RapidFuzz fuzzy matching map ingredient text to USDA FoodData Central records; OpenAI GPT parses free-text meal descriptions into structured items and quantities.
   • Data normalization: combined model outputs with USDA datasets to normalize portion sizes, convert units and map macronutrients accurately.
   • Backend: designed the Python backend for image processing, model inference and nutrition computation, with secure REST APIs using JWT authentication, middleware validation and rate limiting.
   • Performance: exported models to ONNX Runtime and loaded them once at startup, resized and compressed images on-device before upload, moved inference to an asynchronous worker queue and cached nutrition lookups in Redis, reducing p95 latency by 40%.
   • Mobile: enhanced the React Native app with camera capture, barcode scanning and asynchronous processing, increasing successful analyses by 30%.
3. Dec 2020 – Dec 2022

   Full-Stack Software Engineer

   Shiv Infotech

   Designed and delivered six production full-stack platforms, including an HRMS and payroll portal, a revenue-based financing system and a real-time logistics platform, from architecture to deployment, while mentoring a team of ten developers.

   • 6production apps
   • −45%query latency
   • −40%deployment failures
   • 95%code coverage

   • Angular
   • TypeScript
   • Next.js
   • Node.js
   • Sequelize
   • MongoDB
   • Docker
   • Kubernetes
   • Kafka

   Full scope of work

   • Delivery: designed and delivered six production-ready full-stack applications with Next.js, Angular (TypeScript) and Node.js, owning architecture, development and deployment.
   • Data layer: modeled relational data with Sequelize ORM on PostgreSQL and MySQL (migrations, associations, validation) and document data with MongoDB and Mongoose.
   • HRMS portal: built payroll, salary structures, leave and holiday request workflows and payslips in Angular and Node.js, with PDF and CSV export for employees and HR.
   • DevOps: containerized services with Docker, deployed them on Kubernetes and automated CI/CD with GitHub Actions, reducing deployment failures by 40%.
   • Performance: integrated Redis caching and advanced SQL stored procedures, reducing query latency by 45%; improved API performance by 40% with lazy loading and Webpack optimizations.
   • Streaming: built a Kafka-based streaming architecture for distributed services, enabling ~200ms logistics updates.
   • Quality: unit and end-to-end testing with Mocha and Jest to 95% coverage, cutting post-deployment bugs by 30%.
   • Leadership: mentored ten developers through code reviews, debugging and sprint planning to raise coding standards and delivery speed.

Academic foundation & campus leadership

Graduate training in AI, data and systems, alongside leadership roles on campus.

Selected builds

Production systems, graduate work and products in active development. Open any project for the problem, the architecture and the results.

Technical depth, layer by layer

The complete toolkit from my resume, grouped by where each piece sits in a production system.

Certifications & continuing education

Programs completed alongside my degrees, most recent first.

How I work with people

Shipping software is a team effort. These are the habits I bring to the teams and communities I'm part of.

Let's build something that ships

I'm open to Full-Stack, Frontend, Backend and AI Engineer roles, as well as contract work. Messages sent here go directly to my inbox.