Jakka Anjali
Student
Final-year CSE student at VIT Vellore skilled in Python, Java, AI/ML, data analytics, and RAG, passionate about building practical AI-driven applications.
Skills & Tools
- Java
- MySQL
- GitHub
- HTML
- CSS
- Data Structures and AlgorithmsData Structures and Algorithms
- Object Oriented ProgrammingObject Oriented Programming
- CS FundamentalsCS Fundamentals
- Python
- AWS
- NumPy
- Pandas
About Me
Final-year B.Tech Computer Science student at VIT Vellore with experience developing AI/ML applications, query-driven analytical systems, LLM-powered document question-answering solutions and developing Arduino- and ESP32-based prototypes, sensor-driven applications, and MQTT-connected monitoring systems. Knowledge of IoT architecture, device communication, embedded systems concepts, edge/fog computing, and IoT security and privacy. Skilled in Python, Java, data processing, anomaly detection, and integrating machine learning models into applications. Interested in building practical AI solutions combining intelligent data processing, retrieval-augmented generation, and software engineering.
Education
2023-2027
B. Tech in Computer Science and Engineering
Vellore Institute of Technology, (VIT, Vellore)
Selected Work
NoteWitch– Offline RAG PDF Question Answering System
Developed a 100% offline document question-answering system using Retrieval-Augmented Generation (RAG) to query PDFs with zero cloud exposure. Implemented sliding-window text chunking via PyMuPDF, generated 384-dimensional dense embeddings using Sentence Transformers on PyTorch, and built a FAISS vector index for sub-millisecond semantic similarity search. Integrated a local Mistral 7B model through Ollama with real-time token streaming and session caching to synthesize accurate, context-grounded responses.
Parkinson’s Edge– IoT Movement Monitoring Wristband
Engineered an edge-computing wearable prototype using an ESP32 microcontroller and a 6-axis MPU6050 IMU for real-time motion capture. Implemented on-device signal processing routines using 50-sample sliding windows to extract kinetic motion features and distinguish stationary, normal, tremor-like, and fall-like patterns. Applied lightweight rule-based classification directly on the microcontroller for low-latency detection, with MQTT telemetric streaming configured over a Mosquitto broker for remote health monitoring.
FinTrace– Agentic AML Investigation System
Developed a query-driven financial transaction analysis system that routes analyst queries to relevant workflows, including transaction filtering, anomaly detection, risk classification, and customer-level analysis. Applied Pandas for data preprocessing and feature engineering and Scikit-learn’s Isolation Forest for unsupervised anomaly detection.
CAT Smart Operator Assistant– Heavy Machinery In-Cab Co-Pilot
Developed an industrial telemetry assistance platform utilizing simulated heavy-equipment sensor streams. Integrated real-time MQTT message brokers to ingest high-frequency machine operational data, building machine learning-based task-time prediction models and anomaly classification pipelines. Combined predictive analytics with rule-based safety threshold logic and constructed an interactive Streamlit dashboard for live operator safety alerts and equipment diagnostics.