nishchalpr@ai:~$ portfolio
open to roles
nishchalpr@ai:~$ cat whoami.md

Nishchal P R.

>

Building LLM-powered retrieval and automation systems for large-scale unstructured data.

view projects get in touch github
LLM
pipelines
GraphRAG
retrieval
ReAct
agents
Knowledge
graphs
knowledge_graph.live · 36 nodes
drag · hover · click nodes
//about
bio.txt

I build production-oriented AI systems where retrieval, reasoning, and real-world data meet. My focus is LLM-powered pipelines, GraphRAG, and autonomous agents — turning messy, unstructured corpora into structured knowledge and acting on it. I care about the boring parts: latency, precision, evals, tradeoffs.

Currently shipping investment-intelligence retrieval at Naples Ventures and finishing my B.E. in ECE at NIE Mysore.

now.log
AI Intern
Naples Ventures
Jan 2026 — Apr 2026
loc
India
IST (UTC+5:30) · remote-friendly
stats.json
1M→10K
nodes compressed
high-signal graph reduction
2
production AI systems
GraphRAG + autonomous agent
ReAct
agent orchestration
reason + act loops
Multi-hop
graph reasoning
beyond vector search
stack.yaml
languages
PythonC++SQL
frameworks
FastAPILangChainLangGraphLlamaIndexLLM APIs
ai & retrieval
RAGGraphRAGKnowledge GraphsEmbeddingsSemantic Search
tools
Neo4jDockerSupabaseGitNode.jsn8n
concepts
Retrieval SystemsMulti-hop ReasoningEntity LinkingPrompt Engineering
//projects
proj_01

Graph RAG for Investment Intelligence

Graph-augmented retrieval over financial corpora for contextual reasoning beyond vector search.

PythonNeo4jLangChainLLM APIsFastAPI
problem

Vector search alone collapses on multi-hop financial questions where the answer lives in relationships between entities, not in any single chunk.

approach
  • ›LLM-driven entity-relation extraction pipelines for dynamic knowledge graph construction
  • ›Hybrid retrieval combining semantic similarity with graph traversal for contextual grounding
  • ›Graph reduction & ranking pipelines compressing ~1M nodes down to ~10K high-signal nodes
  • ›Mitigation of entity duplication, noisy relations, and signal loss during pruning
wins
  • Multi-hop reasoning unlocked on questions vector RAG could not answer
  • ~100× graph compression with preserved retrieval precision
  • Latency-stable retrieval under iterative graph updates
proj_02

Autonomous AI Logistics Agent

ReAct agent that investigates operational logistics data and executes multi-step issue resolution.

PythonFastAPILangGraphLLM APIsDocker
problem

Support teams burn hours stitching together shipment events, logs, and tickets to root-cause a single delayed delivery.

approach
  • ›Agentic tool-calling with a hybrid retrieval framework over shipment events, logs, and support records
  • ›ReAct (Reason + Act) orchestration loop for autonomous anomaly diagnosis and RCA
  • ›Evaluation pipeline benchmarking tool-selection accuracy, reasoning trajectory, and latency
wins
  • Autonomous root-cause analysis from raw operational signals
  • Tool-selection accuracy measured & tracked, not vibes
  • Production-oriented latency and reliability targets
//experience & education
  1. Jan 2026 — Apr 2026 [current]

    AI Intern @ Naples Ventures

    Fintech B2B SaaS · Investment Intelligence Systems
    • ›Built LLM-powered pipelines processing financial data into structured knowledge graphs
    • ›Worked cross-functionally on retrieval optimization and AI workflow improvements
    • ›Owned graph-based retrieval and contextual information extraction
    • ›Made tradeoffs across scalability, retrieval precision, and latency
  2. 2022 — 2026

    B.E. Electronics & Communication @ National Institute of Engineering, Mysore

    Bachelor of Engineering
//contact

Let's build something
retrieval-aware.

Open to AI engineer roles, GraphRAG / agent contracts, and interesting research collaborations. Fastest reply via email.

$ cat email
nishchalpr4@gmail.com
$ open linkedin
https://www.linkedin.com/in/nishchalpr/
$ open github
https://github.com/Nishchalpr4