class Abhay:
roles = ["Data Analyst", "Data Scientist",
"AI Engineer", "ML Engineer"]
education = "B.Tech, Electronics & Computer Engg (2026)"
college = "Jawaharlal Nehru Engineering College"
cgpa = 7.52
base = "Hyderabad, India | PAN India / Remote"
learning = ["Agentic AI", "LangGraph", "MCP",
"Tool Calling", "Agent Evaluation"]
certifications = [
"OCI 2025 Certified AI Foundations Associate",
"Complete Data Analyst Bootcamp (Udemy)",
"Python: Predictive Analytics (LinkedIn Learning)",
]
principles = [
"Question the data before modeling it",
"Benchmark many models, keep the honest winner",
"Document what failed, not just what worked",
"Ship every project with an API + a UI",
]
def open_to(self):
return "Full-time roles & internships 🚀" |
|
Find the story in the data
Clean messy tables, explore patterns, build dashboards and clear visuals, and turn numbers into decisions. Proof → Retail Sales EDA, ABC Tech ITSM incident analytics |
Model it, test it, trust it
Feature engineering, honest benchmarking, leakage-safe splits and forecasting. Proof → Home Loan Default, ABC Tech ITSM |
Ship it and make it useful
APIs, dashboards, vision pipelines and LLM apps with retrieval. Proof → AI Documents Chatbot, Smart Proctoring, News Clustering |
Languages, Data & Tools
Machine Learning & Deep Learning
Backend & Apps
Analytics & BI
Cloud & DevOps (foundational)
📚 Skills by domain (click to expand)
| Domain | What I work with |
|---|---|
| 📊 Data Analysis | SQL (MySQL), Excel, Power BI, Tableau, data cleaning, EDA, Pandas, NumPy, Matplotlib, Seaborn, Plotly, turning findings into clear visuals |
| 🔢 Classical ML | Feature engineering, imbalanced data (SMOTE), Optuna tuning, GridSearchCV / RandomizedSearchCV, CatBoost, XGBoost, Ridge |
| 🧠 Deep Learning | CNNs, transfer learning (MobileNetV2, EfficientNetB0), ANNs, TensorFlow / Keras |
| 👁️ Computer Vision | OpenCV, YOLOv8 custom training, Haar cascades, real-time video pipelines |
| 💬 NLP & GenAI | Sentence embeddings, UMAP + HDBSCAN clustering, BART / Gemini summarization, RAG, LangChain, ChromaDB, Groq |
| 📉 Time Series | ARIMA, SARIMAX, Prophet, Holt-Winters, chronological validation |
| 🚢 Deployment | FastAPI endpoints, Streamlit dashboards, joblib / Keras model artifacts |
| ☁️ Cloud & DevOps | Foundational knowledge of Docker, Kubernetes and AWS |
|
Chat with any document
Answers questions only from the uploaded file. Supports PDF, TXT, DOCX, CSV, XLSX, PPTX and HTML with MMR retrieval. |
Real-time cheating detection
Custom YOLOv8n flags phones, books, headphones and extra people. mAP50 0.931 · Recall 0.932 |
Articles → events → timelines
209,508 articles grouped into 4,820 events, summarized with BART / Gemini. |
| Project | What it does | Stack | Headline result | |
|---|---|---|---|---|
| 🏦 | Home Loan Default Predictor | Scores default risk (Low / Medium / High) for ~300K applicants from 7 relational tables, 209 engineered features | CatBoost Optuna FastAPI Streamlit |
ROC-AUC 0.7907 |
| 🎫 | ABC Tech ITSM Analytics | Ticket priority classification, daily volume forecasting, RFC risk prediction on 46K+ incidents | CatBoost SARIMAX FastAPI |
Priority ROC-AUC 0.9414 |
| Project | The question | Tools | |
|---|---|---|---|
| 🛒 | Retail Sales EDA | What do 2022–2024 global retail sales reveal about products, regions and trends? | Python Pandas Matplotlib Seaborn |
flowchart LR
subgraph A["📊 ANALYST"]
direction TB
A1[❓ Ask the question] --> A2[🧹 Clean & validate]
A2 --> A3[🔍 Explore & visualize]
end
subgraph B["🔬 SCIENTIST"]
direction TB
B1[⚙️ Engineer features] --> B2[🏆 Benchmark models]
B2 --> B3[🎛️ Tune & validate]
end
subgraph C["🤖 ENGINEER"]
direction TB
C1[💾 Save artifacts] --> C2[⚡ FastAPI backend]
C2 --> C3[🖥️ Streamlit UI]
end
A --> B --> C
style A fill:#0f2027,stroke:#2DD4BF,color:#fff
style B fill:#132f3a,stroke:#38BDF8,color:#fff
style C fill:#1b2440,stroke:#A78BFA,color:#fff
I'm looking for Data Analyst, Data Scientist, AI Engineer and ML Engineer roles, in India or remote. If you're working on something interesting with data or LLMs, say hi.
"The question of whether machines can think is about as relevant as the question of whether submarines can swim."Edsger W. Dijkstra |