AI & Data Science Generalist with End-to-End Systems

Fatima Firdouse

# *****************@*****.*** +91-918**-***** ï LinkedIn § GitHub + Hyderabad, India

Summary

AI and Data Science enthusiast with hands-on experience in solving real-world problems using data, building end-to-end

systems involving data pipelines, machine learning, and user-focused applications. Worked with large-scale datasets (500k+

records), built bias detection and matching systems, and deployed real-time applications. Comfortable working in ambiguous

environments, breaking down problems, and focusing on impact over complexity.

Skills

Programming & Analysis Python(OOP), Pandas, NumPy,Seaborn, Matplotlib, EDA, Feature Engineering

Machine Learning Scikit-learn, Supervised & Unsupuervised Algorithms, Model Evaluation

NLP & Information Retrieval TF-IDF, Cosine Similarity, BM25, Semantic Search, LangChain, RAG Pipelines

Data & Vector Stores MySQL, ChromaDB, Hybrid Search(BM25 + Semantic), Cross-Encoder Reranking

Deployment & Cloud AWS EC2, FastAPI, Streamlit, Git & GitHub

Observability & Evaluation Langfuse (LLM Tracing), Ragas, Jupyter, HuggingFace, Groq (Llama 3.1 8B)

Experience

Data Science Job Simulation Jan 2026

Boston Consulting Group (BCG) — Forage

• Designed and automated an ETL workflow on AWS EC2 using Python and Pandas/NumPy to ingest and transform

500k+ customer records, cutting preprocessing time by 40% and ensuring reproducible data pipelines.

• Engineered domain-specific features and trained a Random Forest churn model — isolating the top 10 churn

predictors and enabling a 15% increase in targeted retention actions.

• Synthesised model outputs into a concise executive summary for senior non-technical stakeholders, demonstrating the

ability to translate complex ML findings into actionable business decisions.

Projects

HireIQ — AI-Based Ranking and Bias Analysis Platform Live § GitHub Apr 2026

• Engineered a dual-flow ETL/ML inference pipeline in Python that parses data, scores candidate-job matches (0-100),

and runs bias detection across recruiter and candidaye workflows.

• Built a FastAPI backend on AWS EC2 (5+ endpoints), added Langfuse observability, and delivered Streamlit

dashboards for client.

• Implemented context-aware bias detection using LLM prompting, generating explainable audit reports for each job

description.

DocMind — Production ML Retrieval Pipeline with Observability Live § GitHub Mar 2026

• Architected a multi-stage retrieval pipeline using LangChain and Python, orchestrating data ingestion, transformation,

and a 3-stage ranking system (BM25 keyword + semantic vector search fused via RRF, re-scored with cross-encoder

reranking) — achieving a Ragas Context Precision score of 1.0 (maximum possible) on automated evaluation.

• Deployed the complete system as a publicly accessible live app on AWS EC2, serving real document queries at

sub-2s response time across 5+ FastAPI REST endpoints.

• Integrated Langfuse tracing to capture every retrieval call, generation step, and latency spike on a live dashboard —

reducing debugging time by 40% and demonstrating production-grade observability practices.

Education

B.Tech in Artificial Intelligence and Data Science (Pursuing) Nov 2022 – Jun 2026

Dr. VRK Women’s College of Engineering & Technology, Telangana

Contact this candidate