AI/ML Full-Stack Engineer

ASHISH YADAV

Full-Stack Software Engineer RAG · LLM · AI Engineering · Next.js · Django · Go · AWS · Docker Gurugram, India ************@*****.*** +91-936******* github.com/yashish13579-bit SUMMARY

Full-Stack Software Engineer with 2 years of experience building and running web products at an early-stage startup. Took a customer-facing CV builder from an empty repo to launch, and built one CRM used by five business teams. Comfortable across the stack: Next.js and React on the front end, Django, Flask and Go on the back end, MySQL/MongoDB, Docker and AWS EC2 in production. Handled sprint planning, code reviews and delivery for a small team, and mentored two junior developers.

Also experienced in Machine Learning, Deep Learning, NLP, Computer Vision, RAG and LLM-based applications — roughly 1 year hands-on with retrieval-augmented generation, document/context retrieval and LLM API integration, plus LLM fine- tuning workflows and prompt engineering. Earlier worked on training and deploying ML models with PyTorch, Flask and Docker.

EXPERIENCE

Full Stack Developer and AI Engineer Nov 2025 – Jul 2026 Defitex Innovative Solutions Pvt Ltd · Gurugram, India In-house CRM — enterprise workflow system

• Before this, Recruiters, Finance, Ops, HR, and the CEO's office were each tracking their work in a different spreadsheet or tool, which made cross-team reporting a manual, error-prone process. I designed a single CRM that gives each team its own workflow — recruiter pipelines, finance approvals, HR admin — while feeding everything into one reporting layer that leadership actually looks at.

• The tricky part wasn't any one module; it was keeping five very different workflows from turning into five separate apps glued together. I kept the data model unified underneath even though each team's UI looks different, with the front end built as React components inside Next.js.

• Alongside the main Django API, I built a set of supporting Go microservices for the parts that didn't belong in the request/response cycle — a background cron service running scheduled jobs like report rollups and data syncs across the five team workflows, and a scraper that pulled in external data the recruiters were previously copying by hand. Each service exposes its own small REST API so the CRM backend can trigger and monitor jobs, and everything runs concurrently with goroutines, Dockerized on the same EC2 setup as the rest of the stack.

• Built an in-house AI chatbot for the CRM using RAG, letting users retrieve relevant information from internal business context and documents and get contextual answers through an LLM. Tech: Next.js (React), Python, Django REST Framework, Go (Golang), MySQL, Docker, AWS EC2, RAG / LLM APIs CV Builder — new product launch

• This was the company's first customer-facing product, and I was handed it at the architecture stage, before a line of UI code existed. The brief was simple to say and harder to build well: let anyone put together a professional-looking resume, tweak it live, and export it, without the experience feeling like a clunky form.

• I broke the interface into reusable React components early on rather than hard-coding each resume template, which paid off a few weeks in when we needed to add new layouts fast — that decision saved us from a rewrite. Node.js API routes handle the backend logic, MongoDB stores the documents, and the whole thing runs on AWS EC2. Tech: Next.js (React), Node.js, Python, MongoDB, AWS EC2 Client CRM

• Built this for an external client who needed something shaped around how their sales team actually sells, not a generic template. Spent time upfront understanding their existing (very manual) process before writing any code, then built it out with Next.js and React on the front end and a Flask API behind it, with MongoDB handling storage. Tech: Next.js (React), Python, Flask REST API, MongoDB, Railway Company & client websites

• Built and still maintain the company's own site on Next.js (React), hosted on AWS EC2. Separately delivered a client website on WordPress — less exciting technically, but it meant getting comfortable customizing themes and plugins to match a client's exact branding requests, hosted on Hostinger. Tech: Next.js (React), AWS EC2 · WordPress, Hostinger Leading the team

• Once I moved into the tech lead role, my day-to-day shifted from writing every feature myself to making sure the team could. That meant running sprint planning, reviewing code before it shipped, and owning delivery timelines across both in-house and client work at the same time.

• I also became the translation layer between leadership and engineering — taking what the CEO or Chief of Staff wanted in business terms and turning it into something the team could actually scope and build.

• Mentored two junior developers directly and pushed for consistent coding standards across the team instead of letting every project pick its own conventions. Over nine months, kept several projects — internal and client-facing — moving in parallel without missing a deadline.

Machine Learning, Deep Learning & AI Intern May 2024 – Nov 2025 Appwar's Technologies · Noida, Uttar Pradesh

• Most of my work here was less about building models and more about getting them to survive contact with production

— wrapping trained models in Flask REST APIs and containerizing them with Docker so they could actually be called by something other than a Jupyter notebook.

• Built and compared a few supervised models — logistic regression, random forest, and XGBoost — for predicting customer churn, and got to 89% accuracy after enough iteration on feature engineering and cleaning in Pandas and NumPy that I stopped trusting the first version I had built.

• Trained a CNN using transfer learning (ResNet50, VGG16) to flag diabetic retinopathy from retinal scans, which ended up around 40% faster at detection than our baseline model.

• Built a BERT-based support chatbot in PyTorch that cut average response time by roughly 60%. Also spent a fair amount of time on the less glamorous parts — cleaning and labeling text and image datasets, and evaluating results properly with confusion matrices, precision/recall, and A/B tests rather than trusting accuracy alone. SKILLS

Languages: Python, JavaScript, Go, SQL

Frontend: Next.js, React, React Native, HTML, CSS

Backend: Django REST Framework, Flask, Node.js, RESTful APIs Databases: MySQL, PostgreSQL, MongoDB

Cloud & DevOps: AWS (EC2), Docker, Railway, Git, GitHub AI/ML: PyTorch, scikit-learn, XGBoost, NLP, Computer Vision, Deep Learning, CNNs, BERT, Transfer Learning Generative AI: RAG (Retrieval-Augmented Generation), LLM Applications, LLM Training/Fine-tuning, Prompt Engineering AI Tools: Claude Code, OpenAI Codex, Gemini CLI, Cursor, GitHub Copilot, ChatGPT, OpenAI / Anthropic / Gemini APIs EDUCATION

B.Tech, Computer Science Engineering 2019 – 2023

Dr. APJ Abdul Kalam Technical University, Uttar Pradesh

Contact this candidate