Company Description
At Neural City, our vision is to enhance the liveability of Indian cities by revolutionizing city governance through technology. Neural City (www.neuralcity.in) is a bootstrapped startup focused on creating transformative solutions in the B2G (Business to Government) domain. We aim to enhance urban planning, management, and monitoring using cutting-edge technologies.
We are currently seeking a highly motivated Applied Gen AI Intern to join our team and contribute to developing innovative solutions that classify and rate existing city infrastructure and services parameters, such as roads, footpath, cleanliness for smart monitoring systems.
Location: Remote
Duration: 4 to 6 Months
Stipend
Tentatively between ₹10,000 to ₹12,000 per month.
Higher for candidates with suitable job and cultural fit.
What You’ll Work On
At Neural City, we build AI systems for messy, real-world governance problems not chatbot demos.
You’ll work on rapid AI prototyping across urban governance, geospatial systems, infrastructure monitoring, document intelligence, and public-service workflows.
This role is ideal for someone who wants to build things that interact with actual users, government systems, imperfect data, maps, images, PDFs, videos, and operational constraints.
Main Responsibilities
-
Rapid AI Prototyping: Build end-to-end prototypes for government and urban use cases using LLMs, multimodal AI, RAG pipelines, agentic workflows, OCR/document intelligence, and VLMs.
-
Work With Real-World Data: Handle noisy datasets including PDFs, scanned documents, maps, street imagery, GPS traces, videos, spreadsheets, and public datasets instead of benchmark-only toy problems.
-
Problem Discovery & AI Research: Research domains deeply using AI tools, identify operational bottlenecks, translate workflows into AI-assisted systems, and evaluate technical feasibility quickly.
-
AI Stack Selection & Integration: Experiment with emerging AI tooling including Claude API, LangChain, LlamaIndex, vector databases, open-source models, local inference setups, and multimodal pipelines.
-
Geospatial + Vision Workflows: Work on AI systems involving street imagery, infrastructure analysis, GIS layers, visual scoring systems, and city-scale observations.
-
Iteration & Validation: Ship prototypes fast, test with stakeholders, gather feedback, break things, improve them, and document learnings for production readiness.
-
Technical Documentation: Write clean architecture notes, workflows, API documentation, deployment guides, and handoff material for future scaling.
What You’ll Learn
-
Applied AI Beyond Hype: How real AI systems are built under constraints like latency, poor data quality, governance requirements, and operational ambiguity.
-
Multimodal + Agentic Systems: Hands-on exposure to OCR, RAG, VLMs, structured extraction, autonomous workflows, and AI-assisted decision systems.
-
Government + CivicTech Workflows: How AI interacts with compliance, procurement, public systems, auditability, and large-scale operational realities.
-
Full-Stack AI Prototyping: Frontend interfaces, backend orchestration, APIs, vector databases, cloud deployment, and rapid iteration cycles.
-
Startup Execution: How to ship fast without waiting for perfect specs, perfect datasets, or perfect certainty.
-
Thinking, Not Just Coding: How to break down ambiguous problems, ask better questions, and design systems instead of merely calling APIs.
Ideal Candidate
-
Strong Python fundamentals: Comfortable learning new frameworks, debugging independently, and shipping quickly.
-
Curious About Applied AI: You experiment beyond tutorials and are genuinely interested in practical LLM and multimodal systems.
-
Comfortable With Ambiguity: Can work through unclear requirements, changing priorities, and incomplete information.
-
Strong Builder Mindset: You like prototyping, testing ideas, and figuring things out independently.
Bonus (Not Mandatory)
Experience with:
-
LangChain / LlamaIndex
-
OCR or document AI
-
GIS or geospatial data
-
React / Next.js
-
FastAPI
-
Vision-language models
-
Open-source model deployment