Hikmah Technologies
Production AI on your own data.
I'm Arshad Ansari. I build the data platforms and the approval-gated AI workflows that run on them — self-hosted, internals public — instead of the $150–200k senior hire you'd spend six months landing, or an AI agency that can demo a workflow but has never run one against production data.
I run all of it in production myself, on my own data, before any of it touches yours. Three live systems — so you can check the work before you pay for it.
2 audit slots a month · one full-time engagement at a time — currently open·What “fractional” means →·How I ship AI workflows →
Verifiable, not claimed
Data
- 3
- systems running in production
- 4,912
- pipeline partitions at 99.9%
AI
- 40
- automation workflows
- 52
- permission-gated tools
Who this is for
You probably recognise at least two of these
- →You have more data than a spreadsheet holds, and nobody whose actual job it is.
- →The warehouse bill is growing faster than your usage, and no one can say why.
- →Two dashboards answer the same question differently, and both get quoted.
- →Data work keeps landing on a product engineer who did not sign up for it.
- →A pipeline broke last month and you found out from a customer.
- →There are three AI pilots in the building and none of them has survived contact with production.
- →Someone has already wired an LLM to something that matters, and everyone is quietly nervous about it.
- →You need this working in weeks — not after two quarters of hiring.
If none of those land, you probably don't need me yet — and I'd rather say so here than on a call.
Your real alternatives
What you'd do instead
Hire a senior data engineer
$150–200k/yr + 3–6 monthsYou are hiring someone to invent a platform — and if you have no data lead, you cannot evaluate the candidates.
Give it to your product engineers
A quarter of roadmapThey learn data engineering on your production data, and the platform ends up shaped like whoever was free that sprint.
An agency or offshore team
$15–30k/monthYou get throughput, not judgement — and every rotation takes the context with it.
Buy the managed stack
A meter that compoundsFastest way to have a warehouse at all, and the reason most of the bills I get called about look the way they do.
An AI agency or automation shop
A demo, then a handoverThe demos are genuinely good. Ask how many of their workflows have run unattended against production data for a year, and who owns it the week after handover.
Wire n8n or Zapier to ChatGPT yourself
A weekend, then upkeepThe right way to find out in a week whether the idea works at all — and fine until the first silent failure reaches a customer, because nobody owns the retries, the guardrails or the audit trail.
Me, fractionally
$3,000 to startOne senior engineer who builds and runs the whole stack — the pipelines and the AI workflows on top of them — delivered in weeks and handed over documented, so you can hire into a working platform later instead of hiring someone to invent one.
If you genuinely have 40 hours a week of data work, hire the full-timer. I'll tell you that on the call.
Selected Work
Systems I've designed and built
From a UK capital-markets research platform to enterprise compliance — systems where the numbers have to be right.
Capital Markets
UK capital-markets research platform
Architected the core data system behind a UK research platform serving investors across Europe — high-throughput pipelines reconciling financial data across global markets.
Read the case study →Compliance / Internal Audit
Sehati
Built an internal-audit management platform for enterprise compliance teams — planning, risk assessment, fieldwork and findings, with every step recorded and reportable in real time.
Read the case study →The Lab
Live systems that prove the claims
Not a product catalogue — each of these runs its own business on its own domain. Here, each one is evidence for a claim this site makes.
Ansaar
AI-powered Shariah-compliant investing platform
A trading and investment platform for Indian retail investors with a Shariah-compliance focus — equity, crypto and ETF screeners, institutional "whale-watcher" signals, market-regime classification and ML-driven predictions.
What it proves: the data-platform claims — the 4,912 pipeline partitions at 99.9% on this site are this product's backbone.
Domain Posture
Audit-grade domain hygiene reports
Signed, hash-chained PDF evidence packs covering DNS, email authentication (SPF / DMARC / DKIM), TLS, security headers and CORS — defensible reports you can hand to auditors and customers.
What it proves: audit-grade evidence discipline — signed, hash-chained reports, because "trust me" is not a compliance answer.
Quantamental
Global macro & central-bank credibility analytics
A macro-analysis platform scoring central-bank credibility and tracking macro fundamentals across 171 countries — built on ten free public data sources (World Bank, IMF, ILO, BIS, FRED, OECD and more) and refreshed daily.
What it proves: LLM analysis over a governed warehouse, in production, refreshed daily.
KoyraCloud
Open SourceOpen-source self-hosted PaaS for Docker Swarm
An open-source (AGPL-3.0) platform-as-a-service that brings a Vercel/Render-style push-to-deploy experience to your own infrastructure — image builds, auto-TLS subdomains, persistent storage, workers and cron jobs on Docker Swarm.
What it proves: right-sized self-hosted infrastructure — "near-zero marginal cost" is a running PaaS, not a slogan.
Featured · Open Source
MITAEGIS
A self-hosted, flow-first personal AI orchestration platform
A small fleet of named agents run durable workflows over my own data — tasks, money, knowledge, homelab alerts — and ask for a decision only when they actually need one. It is the system I run my own days on, now open source for anyone to fork.
- →Durable, flow-first automation on Temporal
- →One human-in-the-loop primitive for every decision
- →Local-LLM-first, self-hosted, bring-your-own everything
Services
How I can help
Data Platform Audit
A fixed-price teardown of your data stack: 12–20 hours of senior work over one week — a ranked fix-list, a cost model comparing your spend to a lean rebuild, and a small proof-of-concept on your own stack. $3,000, and if day one shows no credible opportunity, you pay nothing.
Data Platform Build
I build the whole path — from your source systems into a warehouse (DuckDB, ClickHouse or Postgres) and out to dashboards your team can trust. $15–40k depending on scope, delivered in weeks — against the $150–200k senior hire you'd otherwise make.
AI Automation Build
I build AI automation that runs in production, not in a demo: approval-gated agents, document-to-table pipelines, and natural-language-to-SQL with guardrails. $10–25k per bounded process, plus a monthly retainer to run and improve it.
About
A bit about me
I'm Arshad — fifteen years building data platforms where the numbers have to be right; most recently the architect of a UK research platform's core data system.
I'm the author of Local-First Analytics (now on Amazon), a Databricks technology partner, and I run Hikmah Technologies as a focused consulting practice — helping companies get their data infrastructure right the first time.
FAQ
Common questions
What does Hikmah Technologies actually do?
One thing: production AI on your own data. I build the data platform — pipelines, a warehouse, dashboards — and the AI workflows that run on top of it, every consequential action gated by a human approval. It is a solo practice run by Arshad Ansari, so you work directly with the person building the system. The products on this site are the lab: live systems that prove the claims, not a separate business.
How do I know you can actually build this?
Because you can check before you hire me. I run three systems in production: AEGIS, an open-source AI automation platform you can read line by line at github.com/hikmahtech/aegis; Ansaar, a trading data platform live at ansaar.in with 4,912 pipeline partitions backfilled at 99.9%; and Quantamentry, a macro-data product at quantamentry.com scoring 171 countries daily. Most consultants show logos — I show you the source.
Do I need to hire a data team?
No — that is the point. I design, build and run the whole stack solo, from ingestion to warehouse to dashboards, and hand it over documented. A senior data engineer is a six-figure annual commitment and months of hiring; a scoped build with me is delivered in weeks.
Why not just hire a full-time data engineer?
If you truly have 40 hours a week of data work and $150–200k plus months to hire, you should. Most teams that call me have neither: the work is lumpy — heavy while the platform goes up, light once it runs. A fractional senior engineer gives you the judgement without the salary — $3–4k a month for one day a week, advisory only — and everything I build is documented and handed over, so you can hire into a working platform later instead of hiring someone to invent one.
How do engagements and pricing work?
Everything is priced in USD, in the open. The Data Platform Audit is $3,000 fixed: a free 30-minute scoping call first, I only recommend the audit after seeing your cost export, and if day one shows no credible opportunity, you pay nothing — 12–20 hours over one week, with the fee credited in full toward any follow-on build. Builds run $15–40k against the $150–200k senior hire; AI workflows are $10–25k per bounded process plus a monthly retainer; fractional advisory is $3–4k a month for one day a week, three-month minimum. Capacity is real, not marketing: two audits a month, one full-time engagement at a time.
Is it risky to let an AI agent take actions in my business?
It is, if the agent has free rein — so I do not build it that way. Every action that matters routes through a human approval gate before it happens, and everything is logged and traceable. AEGIS, my open-source platform, works exactly this way: 52 permission-gated tools, every human handoff a single record you approve or reject. You can read the code.
My cloud-warehouse bill keeps growing. Can you help?
Usually, yes. A lot of warehouse spend comes from paying per query for workloads that fit comfortably in DuckDB, ClickHouse or Postgres on right-sized infrastructure. The Data Platform Audit maps exactly where your money is going and what a leaner setup would cost — that is often where the audit pays for itself.
More buyer-mechanics questions — ownership, timezones, invoicing — are answered on the FAQ page.
Not ready to talk?
Take the audit and run it yourself
The 18 questions I work through in a $3,000 audit are published in full — cost, correctness, reliability, and whether you can safely change any of it. With what a bad answer sounds like for each. No signup, no gate; most teams find something expensive in the first hour. And for the AI half: the 17 questions I ask before an LLM workflow is trusted in production — who can stop it, what happens when it is wrong, what it really costs.
Or just stay in touch
I write up what breaks and what it costs — pipelines, warehouse bills, and the things that only show up in production. A few emails a month.
No sequence, no pitch deck. Reply 'stop' once and you're off — it reaches me, not a queue.
Indian Businesses
Based in Navi Mumbai
I also build for Indian businesses — lead-generation websites and business systems for manufacturers, traders and firms. WhatsApp is the fastest way to reach me.
WhatsApp me · +91 72084 61733Industrial / B2B
VC Solutions
Designed and shipped a fast, SEO-focused marketing and lead-generation website for an MPCB-licensed industrial solvents and scrap-metal trading business in Maharashtra and Gujarat.
Read the case study →Real Estate
Manas Realty
Built a modern marketing website for a real-estate firm operating since 1991 — project showcase, services and enquiry capture, with privacy-friendly analytics.
Read the case study →Need the data platform, and the AI that runs on it?
I take on a small number of projects at a time so I can do them well. The first step is a free 30-minute scoping call — most builds start with a $3,000 fixed-price audit whose fee comes off the build.
Book a free scoping callOr email [email protected]