harsh = {
    "name": "Harsh Patel",
    "education": [
        "M.Sc. Data Science @ DA-IICT, Gandhinagar",
        "B.E. Computer Engineering @ LDRP-ITR, KSV"
    ],

    "current_role":
        "Research Intern — IR & Language Processing Lab, DA-IICT",

    "focus": [
        "Applied AI",
        "AI Systems",
        "Agentic AI",
        "Retrieval-Augmented Generation",
        "Information Retrieval",
        "Decision Intelligence"
    ],

    "research": [
        "CLEF 2026 — FinMMEval",
        "CLEF 2026 — CheckThat"
    ],

    "engineering": [
        "Python",
        "FastAPI",
        "REST APIs",
        "SQL",
        "Docker",
        "Data Pipelines"
    ],

    "building": [
        "Enterprise AI Systems",
        "Agentic RAG Systems",
        "Decision Intelligence",
        "Retrieval & Search Systems"
    ],

    "open_to": [
        "Forward Deployed Engineer",
        "AI Solutions Engineer",
        "Applied AI Engineer",
        "AI / ML Engineer",
        "Solutions Engineering"
    ],

    "gate": "GATE 2024 Qualified — Data Science & AI | AIR 7106 | Score 335"
}

M.Sc. Data Science student and research intern focused on building applied AI systems that connect data, models, retrieval, reasoning, APIs, and real-world workflows. My work spans enterprise decision intelligence, Agentic RAG, information retrieval, data engineering, and AI research.

I am interested in the layer between:

A Business / Research Problem
            ↓
        Data & Evidence
            ↓
      Models / Retrieval
            ↓
       AI Reasoning
            ↓
     Software / APIs
            ↓
       Real Workflow
            ↓
        Measurable Outcome

My goal is to build AI systems that are not only accurate, but also:

• explainable
• testable
• deployable
• traceable
• useful in real workflows

These are the projects that best represent how I approach AI engineering.

Worked on EdgeQuant Agent, an agentic financial decision-making system focused on reasoning over financial information and maintaining decision context across an evaluation workflow.

Focus

Agentic AI Financial Decision Making LLM Reasoning Memory Evaluation

Worked on a multi-stage scientific source retrieval framework for implicit social-media claims.

Pipeline

Query / Claim
     ↓
BM25
     ↓
Dense Retrieval
     ↓
Cross-Encoder Reranking
     ↓
Fusion
     ↓
Scientific Source Ranking

Evaluation focus

MRR@K NDCG@K Recall@K Reranking Multilingual Retrieval

I am particularly interested in systems that combine:

Data
 ↓
Retrieval
 ↓
Models
 ↓
Reasoning
 ↓
Decision / Recommendation
 ↓
APIs
 ↓
Deployment
 ↓
Evaluation

The goal is not simply to train a model.

The goal is to build a usable AI system around the model.

Building systems where predictions become actionable business decisions.

Designing systems where AI components can reason over context and select tools/actions.

Building hybrid retrieval, ranking and evidence-grounded information systems.

Turning messy real-world data into structured, queryable and reusable datasets.

Treating evaluation as a first-class engineering problem rather than only reporting a model metric.

Beyond my flagship projects, I have worked on:

• Transformer architectures implemented from core attention mechanisms
• BERT / Transformer fine-tuning
• Deep learning architecture experimentation
• Information retrieval evaluation
• Hybrid retrieval pipelines
• Explainability with SHAP and Grad-CAM
• Agentic financial decision-making
• Scientific source retrieval
• Retrieval-augmented generation

Machine Learning
Neural Networks
Deep Learning
Natural Language Processing
Information Retrieval
Optimization Techniques
Statistics
Probability
Linear Algebra
Data Structures
Database Management Systems

Applied AI
      ↓
AI Systems
      ↓
Agentic AI / RAG
      ↓
Production-Oriented Engineering
      ↓
Customer-Facing AI Solutions
      ↓
Forward Deployed / AI Solutions Engineering

• Forward Deployed Engineer
• AI Solutions Engineer
• Applied AI Engineer
• AI / ML Engineer
• AI Integration Engineer
• Solutions Engineering
• Research Engineering

  

Building AI systems that move from data → intelligence → action.

"Build systems, not just models."