Hi, I'm Abhishek Kumar (@abhicodes). I bridge business analytics with production machine learning, distributed computing (Apache Spark, SQL, pgvector), and generative AI platforms. I transform ambiguous multi-terabyte data into verified business decisions.
Test interactive ML threshold decision logic, SQL performance queries, and business metrics directly in your browser.
Analysis derived from the Integrated Retail Analytics case study. Shows how targeted markdowns surge weekly volume by +34.2% while maintaining a 19.8% margin baseline over 45 monitored retail department hubs.
Real-world applications tested, benchmarked, and containerized. Every project links directly to verified source code.
Flagship GenAI • Live on Vercel
Enterprise learning analytics & grounded RAG platform. Ingests video transcripts (SRT) and readings into 384-dimensional dense vector embeddings in PostgreSQL pgvector with Gemini 2.5 Flash anti-hallucination guardrails.
Domain LLM • QLoRA
Specialized 1.1-billion parameter LLM fine-tuned for logistics and freight operations using QLoRA (Rank 16) and 4-bit NormalFloat quantization. Combines FAISS vector database for zero-hallucination procedure lookup.
Deep Learning • 85k Records
End-to-end predictive deep learning system for 85,907 customer interactions. Employs 6-layer Deep Artificial Neural Network (TensorFlow/Keras), SMOTE imbalance handling, TF-IDF NLP on feedback remarks, and interactive Streamlit UI.
Big Data • PySpark
Full big data banking pipeline exercising Hadoop HDFS storage, HiveQL warehousing, Spark ML distributed classifiers, and Spark Structured Streaming for simulated real-time transaction monitoring on marketing cohorts.
MLOps • Docker • MLflow
Productionized machine learning pipeline featuring Docker containers, Jenkins automation, MLflow model tracking, and dual models (flight pricing regression + passenger propensity) deployed with Streamlit.
Business Analytics • Forecasting
Store department sales forecasting model analyzing holiday markdown promotional impacts, regional unemployment indexes, and fuel price volatility to optimize replenishment and retail inventory levels.
Exploratory Data Analysis
Granular exploratory analysis on 119k hotel booking records. Discovers cancellation predictors, ADR seasonality spikes, lead-time thresholds, and customer distribution channels.
Computer Vision • CNN
Deep convolutional neural network model trained to detect facial boundaries using Haar Cascades and classify human emotional states into 7 affective classes in real-time camera feeds.
Time Series • Regression
Regression modeling forecasting hourly urban bike rentals based on solar radiation, temperature, rainfall, humidity, and public holiday calendar cycles.
Natural Language Processing
NLP text classification pipeline evaluating global pandemic sentiment trends. Utilizes regex text cleaning, stopword filtration, n-gram vectorization, and multi-class classification.
Unsupervised Clustering
Comprehensive exploratory segmentation on the global Netflix media library. Evaluates director co-occurrence clusters, genre migration trends, maturity ratings, and international content growth.
Full spectrum analytical skills spanning statistical modeling, database architectures, machine learning frameworks, and production operations.
How I architect robust end-to-end data systems from raw telemetry to low-latency edge deployment.
Deterministic parsing of multi-modal files (SRT transcripts, CSV/Parquet, transactional logs). Handling high sparsity, timestamp synchronization, and schema verification.
Lagged feature engineering, SMOTE balancing, TF-IDF text features, and dense 384-dimensional vector embeddings indexed with HNSW cosine similarity.
Model training via PySpark ML, TensorFlow ANNs, or QLoRA fine-tuning. Strict validation using ROC-AUC, cross-validation, and EvidenceValidator anti-hallucination guardrails.
Automated testing via 97+ Pytest specs, containerization with Docker, and sub-300ms edge routing deployed globally on Vercel and cloud backends.
Data Scientist & Analytics Specialist
Delhi, IndiaI am a Data Scientist and Analytics Specialist with hands-on expertise building production systems across e-commerce, banking, logistics, and ed-tech. My technical focus spans statistical hypothesis testing, distributed PySpark pipelines, and domain-specific Large Language Models.
Recently, I served as the Testing, Integration, and Deployment Lead for the Coursera Multimodal Intelligence Platform, engineering comprehensive 97-spec automated Pytest suites and deploying the live system on Vercel Global Edge CDN.
Whether you are recruiting for a Data Scientist, Senior Data Analyst, or Machine Learning Engineer role, or looking for an analytics collaboration, my inbox is open.
Tagline