ML Engineer with less than a year in GenAI & LLMs
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI & Data Science graduate with hands-on experience building and deploying Generative AI solutions using LLMs and RAG pipelines. Developed a production-grade DR-RAG system with FAISS and HuggingFace Transformers. Proficient in Python, NLP, and API-based deployment, with growing expertise in LangChain and MLOps. Seeking a Python ML Engineer role in Data & Analytics.
VESIT, Mumbai
B.E. · Artificial Intelligence & Data Science
August 1, 2021 – June 30, 2025
GenAI Question Paper Generation System
January 1, 2025 – June 1, 2026
• Designed and deployed an end-to-end GenAI solution using LLMs to auto-generate exam-ready question papers — MCQs, short-answer, essay, and fill-in-the-blank — in under 1 minute for 100–150 questions. • Implemented Dynamic RAG (DR-RAG) with a two-stage FAISS retrieval pipeline: static relevance pass followed by dynamic query refinement, grounding every question in syllabus content. • Engineered semantic deduplication using sentence-transformers, reducing repeated questions by ~40% in internal evaluation. • Built REST API backend with Flask; full-stack deployment (Flask + ReactJS + MySQL) with RBAC and OAuth2, validated for 100+ concurrent users. • Integrated Bloom's Taxonomy (6 levels) and OBE framework; multilingual support (Hindi + English) across CBSE, ICSE, and state board curricula. • Outcome: teacher paper-preparation time reduced from 8–10 hours to under 2 minutes.
Product Review Sentiment Analysis Dashboard
January 1, 2024 – December 31, 2024
• Built an NLP pipeline classifying 50,000+ Amazon reviews as Positive, Negative, or Neutral using pre-trained DistilBERT — no fine-tuning overhead. • Designed interactive Streamlit dashboard: sentiment by category, rating-vs-sentiment agreement, and top keywords driving negative reviews. • Key insight: 23% of 4-star reviews carried negative sentiment language, demonstrating that star ratings alone are an unreliable signal. • Live text input with instant confidence scoring; deployed on Streamlit Community Cloud.
Retail Sales Forecasting Dashboard
January 1, 2023 – December 31, 2023
• Built an end-to-end ML pipeline on Kaggle Superstore data (4 years, 9,994 orders) — cleaning, feature engineering, modelling, evaluation, and deployment. • Structured model comparison: Prophet outperformed ARIMA by 12% MAE, better handling November–December seasonal spikes. • Streamlit dashboard with category selector and 1–12 month forecast horizon; confidence intervals plotted against historical actuals.
Student Performance Predictor
January 1, 2022 – December 31, 2022
• Trained Random Forest classifier on UCI Student Performance dataset (649 students, 33 features); 88% accuracy, F1-score 0.85 — evaluated with confusion matrix, precision/recall, and cross-validation. • EDA identified study time and prior failures as top predictors; feature importance plots used to communicate findings to non-technical stakeholders. • Streamlit app with teacher-facing input form returning prediction + top 3 contributing factors.
Cultural Fit Analysis
The candidate's projects demonstrate a strong interest in applying AI/ML to solve real-world problems, aligning with an innovative and problem-solving culture. The diversity of projects (GenAI, NLP, forecasting, classification) shows a broad interest in different ML domains. However, all projects are academic or personal, and there is no professional experience, which might indicate a need for mentorship in a corporate environment.
Soft Skills & Operational Fit
The candidate's role as a Class Representative suggests good communication, coordination, and stakeholder management skills. The project descriptions indicate an ability to communicate technical findings to non-technical stakeholders (e.g., feature importance plots, key insights from sentiment analysis).