AI / ML · Full Stack · ERP · Field notes
Build logs from a working AI/ML & Full Stack Developer — shipped in production, written without the marketing.
Field notes from client work: RAG vs fine-tuning, cutting an OpenAI bill without breaking quality, Retell voice agents, Next.js dashboards, slow ERP queries, and deploy mistakes I have already paid for.
RAG vs Fine-tuning: How I Actually Decide for Production AI Apps
A practical decision framework from a working AI developer — when retrieval beats fine-tuning, when it does not, and the cost numbers nobody puts in their slides.
Read the articleCapabilities
What I build — hire an AI/ML and full stack developer in India
LLM apps, RAG pipelines, AI dashboards
Production-grade AI features wired into your real data and workflows — OpenAI, Anthropic, Retell, embeddings, evaluation harnesses, cost controls. Not demos.
Retell AI & LLM voice ops
Inbound voice agents that qualify leads, take bookings, and escalate cleanly to humans — with the latency tuning that makes the conversation feel real.
Performance & reliability
PHP, Node, and cloud-native patterns: caching, queues, clean schemas, and deployments that survive Monday-morning traffic spikes.
Internal platforms that ship
Modules for HRMS, payroll, finance, inventory, and approvals — built so non-developer teams can run day-to-day work without breaking production.
Articles & field notes
Long-form notes from real production work.
Each post draws on an actual client engagement — the architecture choices, the numbers, and the bits that did not work the first time.
How I Cut a Client's OpenAI Bill by 62% Without Hurting Quality
Eight techniques that reduced a production AI app's monthly inference bill from $1,470 to $558 — model routing, prompt diet, caching, and the boring infra wins nobody talks about.
Building a Production AI Dashboard with Next.js, Node, and Streaming LLMs
A field guide from the dashboard I ship at Technosmart — App Router, server actions, SSE streaming, Retell webhooks, and the small architecture choices that decide whether your AI product feels real.
Claude vs GPT-4 in Production: A Working Developer's Take
I ship features on both. Here is how I actually choose between Claude and GPT-4 for real products — broken down by task, latency, tool use, and the small behaviours that decide it.
Integrating Retell AI for Voice Ops: A Production Walkthrough
Building voice agents that actually answer real customer calls — webhook handling, prompt design, escalation logic, and the latency tricks that keep conversations feeling human.
Why OOP PHP Still Quietly Wins for ERP — Lessons from 9,000+ Daily Users
Object-oriented PHP is unfashionable. It is also still the best fit for ERP work where modules outlive trends and every change has to land without breaking the rest of the system.
How I Took an ERP Report from 5 Seconds to 90 Milliseconds
A blow-by-blow account of one query optimisation engagement — EXPLAIN plans, the composite index that did the heavy lifting, and the materialised view that finished the job.
Contact Flows That Actually Convert (Without Resorting to Pop-ups)
Specific UX moves I have used on real client sites to turn a flat "Contact Us" page into a respectable enquiry pipeline — including the rule that doubled the conversion rate on one of them.
Local SEO for Small Indian Businesses: A No-Fluff Playbook
The local-SEO moves that actually move rankings for small service businesses in India — Google Business Profile, schema, page speed, and the bits that finally got my Australian client to rank in Canberra.
The Zero-Panic Deployment Checklist I Actually Use
A printable pre-launch checklist that has saved me four embarrassing launches in a row — domains, SSL, redirects, caching, forms, error pages, and the boring stuff Google notices.
No articles match that search. Try a different keyword or category.