NIDHEESH TK

Software Engineer

• Developed and maintained FastAPI backend microservices and REST APIs, implementing business logic, request validation, and error handling with PostgreSQL persistence.
• Engineered and executed 70+ test cases across complex real-time pricing algorithm scenarios (OPA), conducting functional, regression, integration, and API testing using Postman and SQL in an Agile framework.
• Resolved production issues by performing root cause analysis across API, service, and database layers.

Software Engineer

• Executed functional, regression, API, and application testing for core software modules.
• Transitioned into backend engineering to deliver Python, FastAPI, and PostgreSQL microservices and production bug fixes.
• Built exposure to LangChain, LangGraph, LLMs, RAG, and Agentic AI workflows to develop AI-based applications.

Software Engineer

• Developed and maintained FastAPI backend microservices and REST APIs, implementing business logic, request validation, and error handling with PostgreSQL persistence.
• Engineered and executed 70+ test cases across complex real-time pricing algorithm scenarios (OPA), conducting functional, regression, integration, and API testing using Postman and SQL in an Agile framework.
• Resolved production issues by performing root cause analysis across API, service, and database layers.

Software Engineer

• Executed functional, regression, API, and application testing for core software modules.
• Transitioned into backend engineering to deliver Python, FastAPI, and PostgreSQL microservices and production bug fixes.
• Built exposure to LangChain, LangGraph, LLMs, RAG, and Agentic AI workflows to develop AI-based applications.

Intern

• Interned in Artificial Intelligence and Machine Learning under Dr.
• Ameer P M.
• Built an EfficientNet-based binary classification model to distinguish benign vs.
• malignant melanoma lesions — 91.45% validation accuracy on a 10K+ image dataset.
• Evaluated model robustness against adversarial attacks (BIM, PGD via Torchattacks), with accuracy dropping to 73.79% under BIM attack, and implemented an SVD-based defense mechanism.

AI/ML Intern – Classification & Adversarial Robustness

• Developed EfficientNet-B0 binary and multiclass melanoma classifiers using PyTorch on datasets up to 35,331 images, achieving up to 91.45% validation accuracy.
• Evaluated classification model vulnerability against BIM and PGD adversarial attacks using Torchattacks.
• Implemented an SVD-based defense mechanism to mitigate accuracy degradation under adversarial conditions.

Intern

Agentic Code Editor

LangGraph, Python, AST, LLM Constructed an interactive CLI code-editing agent utilizing a multi-stage LangGraph pipeline to convert natural-language requirements into functional code modifications.

LangGraph, Python, AST, LLM

RAG Documentation Assistant

FastAPI, Qdrant, Groq, BM25, Cross-Encoder Reranking, Sentence-Transformers, Docker Compose Architected a FastAPI RAG service to ingest documents, compute vector embeddings using Sentence-Transformers, and manage semantic search with a Qdrant vector store.

FastAPI, Qdrant, Groq, BM25, Cross-Encoder Reranking, Sentence-Transformers, Docker Compose

CreditWise – Loan Approval Prediction

Python, scikit-learn, KNN, Logistic Regression, Naive Bayes Built an end-to-end supervised machine learning pipeline to predict loan approvals, benchmarking KNN, Logistic Regression, and Naive Bayes models on financial datasets.

Python, scikit-learn, KNN, Logistic Regression, Naive Bayes