Nivyan.AI Engineer & AWS Cloud Engineer
I build full-stack systems, LLM integrations, and cloud infrastructure that holds up — without paging anyone at 3am.
"focus": ["web", "AI/ML", "cloud"],
"cloud": "AWS Serverless",
"status": true
}
I build things
at the edge of
what's possible.
I've been engineering software for six years — from solo freelance builds to leading delivery on multi-tenant SaaS. The throughline: building systems at the intersection of elegant code and real-world impact.
From LLM-powered tools and Retrieval-Augmented Generation pipelines to distributed AWS serverless backends, I work across the full stack with a bias toward infrastructure that scales — with production experience in React, TypeScript, Next.js, Python, FastAPI, LangChain, and cloud-native patterns built to outlive the demo.
Recent production work spans Promptly (a Socratic AI product architect where 16 metered agents propose and deterministic code decides — a four-gate cost guard, an acyclic dependency graph, a living PRD, all behind FastAPI + Anthropic Claude + Next.js), SBA Loans HQ (a multi-tenant SBA lending SaaS with four portal experiences from one codebase, a blind-listing deal marketplace, AWS Textract + Gemini document automation, and $33M+ in loans processed), and Quanta (a self-hosted AI BI platform with a six-stage natural-language-to-SQL pipeline across four LLM providers, SELECT-only safety enforcement, and SSH bastion tunneling to private databases).
The full toolkit.
hover to explore
Production case studies.
Promptly
Socratic AI product architect: 16 agents propose, deterministic code decides — a four-gate cost guard, an acyclic dependency graph, and a living PRD from one spec store.
SBA Loans HQ
Four-portal SBA lending SaaS: blind deal marketplace, 10-stage enforced pipeline, AI document intake — $33M+ in loans processed.
Quanta
Self-hosted AI BI: ask in plain English, a six-stage pipeline writes provably read-only SQL — even against databases behind SSH bastions.
Where I've shipped.
- 2025 — Present
Full-Stack AI Engineer & Product Manager
SBA Loans HQ · ArithmiksTook dual ownership of engineering and product direction on a multi-tenant SBA lending platform. Shipped AI-powered document processing, financial audit automation, loan lead spam filtering, and approval prediction tools across the full loan lifecycle. Over $33 million in loans approved through the platform, with a substantial pipeline of active deals in progress.
- 2024 — 2025
Full-Stack Developer
Swerv Automotive · ArithmiksBuilt a predictive lead intelligence platform for automotive dealerships. Analysed vehicle lifecycle and purchase history data to surface high-intent buyers before they enter the market, delivering ranked prospect lists with likelihood scores so dealerships can act ahead of demand.
- 2023 — 2024
Front-End Developer
Easybar · ArithmiksDelivered a full-stack rebar procurement platform that replaced fragmented manual workflows with a unified digital system. Built a canvas-based interactive design tool for rebars, wire mesh, and pile cages with real-time visual feedback; automated quotation generation and print-ready fabrication ticket delivery; branded PDF and Excel exports with RTL/LTR language support; centralised order tracking with automatic weight and timeline calculations; and a side-by-side bid comparison tool for procurement decisions.
- 2020 — 2025
Independent Software Consultant
Contract · Remote20+ end-to-end products shipped for global clients across healthcare, e-commerce, CAD tooling, and 3D visualization. Trusted with repeat engagements across every industry, delivering on time and within scope across the board.
Common questions.
- What kind of engineering work do you take on?
- Production AI/LLM features (RAG, agents, structured outputs across Claude / GPT / Gemini), AWS serverless architecture, and full-stack web builds with Next.js, React, FastAPI, and PostgreSQL. Strongest on multi-tenant SaaS, AI-augmented workflows, and cloud-native systems where reliability matters.
- Are you available for full-time roles?
- Yes — open to senior software engineer, staff engineer, or tech lead positions, remote-first, ideally with EMEA or US-East timezone overlap. Also open to fractional and contract engagements.
- Can I hire you as a freelance AI engineer?
- Yes — I take freelance and contract AI engineering work: LLM integration (Claude, GPT, Gemini), RAG pipelines, AI agents, and AWS serverless builds. Engage directly by email, through Contra, or as a fractional senior engineer.
- What's a typical engagement length?
- Contract work usually runs 4–16 weeks for a focused build (LLM feature integration, MVP, or AWS migration), or ongoing as a fractional senior engineer. Full-time roles are open-ended.
- What timezones do you work across?
- Based in Lahore, Pakistan (UTC+5 / Asia-Karachi). Standard overlap with EMEA all day and US-East mornings. Comfortable shifting hours for sync-heavy clients.
- Which LLM and AI tools have you shipped to production?
- Anthropic Claude (Sonnet and Opus), OpenAI GPT-4, Google Gemini, LangChain, embeddings and vector retrieval, AWS Textract for OCR, and custom evals. Recent work: Promptly (a Claude-powered Socratic product architect with a four-gate AI cost guard and 16 metered agent pipelines) and Quanta (natural-language SQL across Gemini, OpenAI, Claude, and Groq with SELECT-only enforcement).
- How do I scope a project quickly?
- Email hello@nivyan.dev with one paragraph on the problem, current stack, and timing — or book a 30-minute call at calendly.com/bnivyan09. I respond within a business day.
Let's build
something
worth shipping.
Available for senior individual contributor roles, cloud architecture consulting, AI/LLM integration work, and 0→1 MVP builds. Remote-first across EMEA and US-East timezones. Inbox is fast.