AI/ML Engineer and Computer Science graduate with hands-on experience building production-grade LLM-powered applications, ML models, and enterprise workflow automations.
- 🤖 Built a Natural Language SQL Chatbot integrating Groq's Llama 3.3 70B, LangChain, and RAG on a live SQL Server backend — achieving 91% semantic SQL accuracy across 500+ test queries
- ⚙️ Engineered an end-to-end GL Entry Approval automation connecting SAP SuccessFactors and Sage X3 via n8n Cloud with OAuth 2.0, audit logging, and error handling for 15+ failure scenarios
- 🚀 Deployed an AI-powered CV Screening App on SAP BTP using SAP CAP, Fiori Elements, Node.js and Groq AI — live on Cloud Foundry
- 📊 Building a Universal AI Data Dashboard (Streamlit + Plotly + Groq) supporting any file format with natural language data querying
- 💡 Passionate about building AI systems that solve real business problems at scale
🔍 Open To: AI/ML Engineer · SAP BTP/CAP Developer · Integration Engineer · Full Stack Developer (AI-focused)
| Domain | Proficiency | Key Tools & Details |
|---|---|---|
| LLM Engineering | ⭐⭐⭐⭐⭐ | Groq API · LangChain · Llama 3.3 70B · Prompt Engineering · Chain-of-Thought |
| RAG Pipelines | ⭐⭐⭐⭐⭐ | FAISS · ChromaDB · Pinecone · Weaviate · Schema-Injected Context |
| AI Agents | ⭐⭐⭐⭐ | SQL Agent · Tool-Use Agents · LangChain Agent Executor |
| Data Visualization | ⭐⭐⭐⭐⭐ | Streamlit · Plotly · Auto-Chart Generation · KPI Detection |
| Computer Vision | ⭐⭐⭐⭐ | YOLOv10 · SegFormer · SAM · Object Detection · Semantic Segmentation |
| Time Series | ⭐⭐⭐⭐ | PatchTST · LSTM · ARIMA · Forecasting · Anomaly Detection |
| Fine-Tuning | ⭐⭐⭐ | QLoRA · PEFT · Hugging Face Transformers |
| ML Ops | ⭐⭐⭐ | MLflow · Experiment Tracking · Model Versioning · FastAPI Serving |
| Workflow Automation | ⭐⭐⭐⭐ | n8n Cloud · OData · REST APIs · OAuth 2.0 · Human-in-the-Loop |
📊 Universal AI Data Dashboard — Streamlit + Groq + Plotly
A fully universal, AI-powered data analytics dashboard that accepts any file format — Excel, CSV, PDF, or Word — and automatically generates visual insights without hardcoding. Powered by Groq's Llama 3.3 70B for natural language data analysis with an intelligent chat interface.
| Attribute | Details |
|---|---|
| Stack | Python · Streamlit · Pandas · Plotly · Groq AI · pdfplumber · python-docx |
| AI Model | Groq llama-3.3-70b-versatile (token-efficient summary-based architecture) |
| File Support | Excel (.xlsx / .xls) · CSV · PDF · Word (.docx) |
| Scale | Handles 10,000+ row datasets via Pandas summarization before LLM call |
| Intelligence | Auto-detects column types, KPIs, date trends, overdue items |
| Security | User-supplied Groq API key via sidebar — zero hardcoded secrets |
| Impact | Universal tool — any industry, any dataset, any team |
| Repository | github.com/akshayy718/universal-ai-dashboard |
Smart token management architecture: Pandas computes a statistical summary first, then only the summary (~300 tokens) is passed to Groq — enabling AI-powered analysis even on 100,000-row files without hitting token limits.
🤖 AI-Powered CV Screening App — SAP CAP + Fiori + Groq (Live on BTP)
Enterprise-grade CV screening application built on SAP CAP, deployed live to SAP BTP Cloud Foundry. Accepts PDF, DOCX, and TXT uploads, extracts structured candidate data using Groq Llama 3.3 70B via a two-stage AI pipeline, and presents results through SAP Fiori Elements UI (List Report + Object Page).
| Attribute | Details |
|---|---|
| Stack | SAP CAP · SAP Fiori Elements · Node.js · Groq AI · SQLite · Cloud Foundry |
| AI Pipeline | Stage 1: Structured JSON extraction → Stage 2: Professional AI summary generation |
| API | OData v4 via CDS service definition |
| Deployment | SAP BTP Cloud Foundry — us10 region — live in production |
| Performance | Extracts name, email, skills, experience, and role fit from any CV format |
| Security | CF environment variable secrets — no hardcoded API keys |
| Repository | github.com/akshayy718/cv-screening-app |
| Live Demo | cv-screening-app.cfapps.us10-001.hana.ondemand.com |
💬 Natural Language SQL Chatbot — LangChain + Groq + MS SQL Server
Production-style AI chatbot converting natural language business questions into validated SQL queries against a Microsoft SQL Server Northwind ERP database. Achieved 91% semantic SQL accuracy across 500+ test queries through systematic prompt engineering.
| Attribute | Details |
|---|---|
| Stack | Python · LangChain · Groq API · MS SQL Server · SQLAlchemy · PyODBC · Gradio |
| AI Model | Groq llama-3.3-70b-versatile via LangChain SQL Agent |
| RAG | Schema-injected context (table names, columns, foreign keys) into prompt |
| Accuracy | 91% semantic SQL accuracy · complex query accuracy improved from 76% → 89% |
| Performance | Query latency reduced from 2.3s → 0.8s after index optimization |
| Safety | Read-only SELECT enforcement — zero destructive query execution |
| Prompt Eng. | 25+ systematic variations · chain-of-thought prompting · 500+ test queries |
| Repository | github.com/akshayy718/northwind-chatbot |
⚙️ GL Entry Approval Automation — SAP SuccessFactors × Sage X3 via n8n
Enterprise workflow automation pipeline on n8n Cloud integrating SAP SuccessFactors (OData GET) with Sage X3 (REST POST) via an email-based human-in-the-loop approval step with full audit logging and OAuth 2.0 security.
| Attribute | Details |
|---|---|
| Stack | n8n Cloud · SAP SuccessFactors OData API · Sage X3 REST API · Gmail API · OAuth 2.0 |
| Pattern | Event-driven · Human-in-the-loop · Multi-day state machine with escalation logic |
| Mapping | 20+ SAP GL fields → Sage X3 journal entry schema (tax codes, contra-account logic) |
| Reliability | Error handling for 15+ failure scenarios · exponential backoff retry |
| Compliance | Full audit logging: approver identity, timestamp, GL details, financial impact |
| Impact | Eliminates manual GL entry routing bottlenecks across two separate ERP systems |
📱 Mobile Device for Impaired Persons — Computer Vision + NLP
Smart Android device for visually, hearing, and speech-impaired users combining real-time object detection, NLP sign interpretation, and emotion-based voice synthesis.
| Attribute | Details |
|---|---|
| Stack | Python · Microsoft Kinect · PCA · Viterbi Search · Android |
| Features | Navigation Detection · Speech Recognition · Face Gesture Recognition |
| AI | Real-time object detection · NLP sign language interpretation · emotion-based TTS |
| Impact | Accessibility tool serving multi-impairment users in a single unified device |
Ramah General Contracting & Transport LLC · Abu Dhabi, UAE · April 2026 – Present
Providing IT support and system coordination for a leading UAE heavy infrastructure and earthworks contractor with 1,300+ employees and a 2.5B+ AED project portfolio — while independently architecting AI and automation solutions to modernise internal workflows.
- Administering Asite (construction project management platform) to maintain structured records of site work report forms, daily progress reports, and field documentation across active construction projects
- Developed a proof-of-concept Natural Language SQL Chatbot (LangChain, Groq Llama 3.3 70B, MS SQL Server) to demonstrate AI-driven data querying for operational reporting use cases
- Architecting an n8n-based GL Entry Approval automation connecting SAP SuccessFactors with Sage X3, eliminating manual finance workflow bottlenecks
- Supporting day-to-day IT operations including user access management, software troubleshooting, and coordination between site teams and management
Corizo · Bengaluru, India · January 2024 – July 2024
- Built Cardiovascular Disease Prediction Model using TensorFlow and Scikit-learn with ensemble classifiers and MLflow experiment tracking
- Gained hands-on experience in Python, ML frameworks, and iterative AI model development workflows
- Trained computer vision models including YOLOv10 and SegFormer for object detection and semantic segmentation
Quantum Gully · 2022 – 2024
- Developed responsive landing pages and web applications for multiple businesses, handling end-to-end client communication and delivery
- Built the RGCT Careers subpage on Wix Studio with CV upload functionality and dark-theme enterprise design
- Developed a Car Price Prediction Flask App with ML model inference and REST API exposure
| 🏅 Recognition | 📋 Details |
|---|---|
| 🎓 B.Tech CGPA 8.01 | Computer Science — AI Specialization · Jain University Bengaluru |
| 🚀 SAP BTP Cloud Foundry Deployment | Independently deployed SAP CAP app to live production environment |
| 🤖 91% SQL Accuracy | LangChain + Groq SQL chatbot across 500+ test queries via chain-of-thought prompting |
| ⚙️ 15+ Failure Scenarios Handled | Robust error handling in n8n GL automation with exponential backoff retry |
| 🔗 Cross-Platform ERP Integration | SAP SuccessFactors ↔ Sage X3 via n8n Cloud with OAuth 2.0 audit trail |
| 📊 Universal AI Dashboard | AI analytics tool supporting Excel · CSV · PDF · Word for any industry |
| 📐 Multi-Domain AI Portfolio | Computer Vision · Time Series · NLP · RAG · Agents · Data Viz · SAP |
current_focus:
learning:
- SAP BTP Advanced: XSUAA auth · SAP HANA Cloud · MTA deployment
- LLM Engineering: QLoRA fine-tuning · RAG pipeline optimization
- n8n: advanced webhook orchestration · complex branching workflows
building:
- Universal AI Data Dashboard → Streamlit + Groq + Plotly
- AI-powered CV Screening App → SAP CAP + Fiori (live on BTP CF)
- GL Entry Approval Automation → SAP SuccessFactors × Sage X3
exploring:
- Agentic AI patterns (tool-use · multi-step reasoning · memory)
- SAP Integration Suite and BTP connectivity services
- Vector database architectures for enterprise knowledge retrieval
open_to:
- AI/ML Engineer → UAE · India · Remote
- SAP BTP / CAP Developer → UAE · India · Remote
- Integration Engineer → SAP · ERP · API platforms
- Full Stack Developer → AI-focused product engineering
"The best engineers don't just write code — they architect solutions that outlive the problem."