AI Engineer with less than a year in LLMs, RAG, and Machine Learning.
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Highly motivated AI/ML Engineer graduating in 2026, with a strong foundation in Artificial Intelligence, Deep Learning, and Natural Language Processing. Proven ability to design and implement complex AI systems, including multi-stage Hybrid RAG pipelines and autonomous agent orchestration. Experienced in fine-tuning large language models, developing perceptual GANs for image super-resolution, and deploying models as web applications, demonstrating a commitment to creating impactful, cutting-edge solutions.
Dhanalakshmi Srinivasan Engineering college
B.Tech · AI and DS
September 1, 2022 – May 1, 2026
Bio-Medical Research System
September 1, 2022 – June 1, 2026
Architected a LangGraph DAG of 11 specialist agents that automates the full biomedical research lifecycle, from parallel literature retrieval across PubMed, ClinicalTrials.gov, and bioRxiv/FDA to hypothesis refinement, protocol design, and a final PI-level GO/NO-GO decision. Engineered a closed-loop validation framework where independent Peer Reviewer and Safety Officer agents trigger up to 3 revision cycles before sign-off, reducing methodological errors and ethical gaps without human intervention. Developed a production-grade backend using FastAPI with Server-Sent Events (SSE) for real-time agent state streaming, paired with a React 19 + TypeScript frontend featuring live progress tracking, evaluation score visualizations, and one-click PDF report export.
View ProjectSRGAN Super-Resolution
September 1, 2022 – June 1, 2026
Re-implemented SRGAN (CVPR 2017) from scratch in PyTorch, training a two-stage pipeline consisting of SRResNet (MSE pre-training, 1M iterations) followed by SRGAN fine-tuning using VGG19 perceptual loss and adversarial loss to achieve 4× single-image super-resolution. Achieved state-of-the-art benchmark performance with 31.93 dB PSNR / 0.902 SSIM on Set5 and 28.59 dB PSNR / 0.799 SSIM on Set14, matching or exceeding the original paper's reported results across standard evaluation datasets including Set5, Set14, and BSD100. Deployed the model as a Streamlit web application for interactive inference and implemented a YAML-driven configuration system to separate hyperparameters, dataset paths, and model checkpoints, ensuring reproducible experimentation and streamlined model management.
View ProjectIndian Legal AI
September 1, 2022 – June 1, 2026
Solved the hallucination problem in legal AI by building a multi-stage Hybrid RAG pipeline, combining BGE-M3 dense + sparse retrieval fused via Reciprocal Rank Fusion (RRF) with a Cross-Encoder reranker, grounding every answer in verified Indian statutes and Supreme Court case law. Fine-tuned Llama 3.1 8B on Indian legal Q&A using QLoRA (4-bit) via Unsloth, achieving 2× faster training and 70% lower memory usage on a single Tesla T4 GPU; exported the model as a merged GGUF for private local inference. Indexed 1,000+ Supreme Court judgments (1950–2025) and 10+ statutes (IPC, CrPC, CPC) into a Qdrant vector database; implemented an LLM-as-a-Judge evaluation framework measuring Faithfulness, Answer Relevancy, and Context Relevance to quantitatively evaluate retrieval performance.
View ProjectCultural Fit Analysis
The candidate's academic projects demonstrate a strong alignment with an AI Engineer role, showcasing diverse applications of AI from biomedical research to legal tech and computer vision. The breadth of technologies used (LangGraph, Qdrant, FastAPI, PyTorch, Streamlit, QLoRA, PEFT) indicates a willingness to learn and apply various tools, which is positive for cultural fit in an innovative environment. The focus on solving real-world problems (hallucination, research automation) suggests a practical and impact-driven mindset. The lack of professional experience means cultural fit is primarily inferred from project scope and technical choices.
Soft Skills & Operational Fit
The candidate's project descriptions indicate a strong problem-solving orientation, particularly in addressing challenges like LLM hallucination and automating complex workflows. The detailed descriptions suggest an ability to articulate technical solutions clearly. The use of closed-loop validation and evaluation frameworks points to a methodical and quality-focused approach. However, without direct interaction or psychometric test results, it's difficult to fully assess stress handling, team collaboration, or broader work attitude.