OPENING CH.1 ORIGINS CH.2 THE OFFICE CH.3 THE DESK CH.4 THE MILL THE BOOKS
Ajay Sachdev presents
A finance and operations manager's statement of work, in four chapters and one schedule

The best problems live in the gaps.

I run finance and operations by day. By night I build the systems the numbers run on. Everything you're about to scroll through is real, in production, and verifiable. Three of them, you can play with.

Scroll to begin the audit
In a hurry? Jump straight to the schedule
OPERATOR: A. SACHDEV BASE: CAMBRIDGE, UK SYSTEMS: 5 PUBLIC / BUILT SOLO
01 / 04 ORIGINS

It all starts with a family business.

Before the spreadsheets and the AI agents, there was a family textile business in Mumbai, India. The route from there to here draws itself as you scroll.

N SURVEYED 2015 - 2026 19.0N / 72.8E 52.2N / 0.1E STUDY WORK BOTH AT ONCE BSc Financial Markets 2015-18 / JAI HIND, MUMBAI Family business 2018-20 / MUMBAI Chartered Wealth Manager 2019 / M.COM (Y1), DIPLOMA MSc Finance 2020-21 / LOUGHBOROUGH, UK Finance & Operations 2021-NOW / 3 COUNTRIES CIMA finalist 2025-26 / STRATEGIC CASE STUDY Cambridge, now BUILDS THE SYSTEMS commercial instinct came before the qualifications did
02 / 04 ASSIST

Ask my other me.

Nobody could find anything in 6,124+ files, so I built OpsPilot: an AI that lives in Teams and answers before I do. Send the message. Watch it work.

OPSPILOT / MICROSOFT TEAMSPRODUCTION / 24/7
Colleague / 09:14Hey Ajay, where's the signed Acme master agreement? Client call in 20 minutes.
OpsPilot / draft for approval Found it: Acme_MSA_signed.pdf in Contracts/2024, countersigned 12 Mar.
Drafted reply with the file link and the two clauses they'll ask about.
Pipeline / live
Triage: work / personal / noise
Intent: document search
Hybrid search: 0 docs scanned
Rerank: top 3 candidates
Draft: Mini-Ajay voice, sensitivity check
100% search accuracy on a 50-query test suite. Frozen until it misses.28 AI tools / 6,124+ docs
03 / 04 RESTRAINT

Try to make my mistake.

I kept making emotional trades, so I built four AI agents to stop me. Below is a real setup from the bot's watchlist. Go on. Place the trade.

ALPHA-PRESERVE / FTSE 100 INTRADAYPAPER MODE / 10:35 GMT
BARC.L / Barclays PLC
price 302.4 / RSI 63.8 / above EMA-50
momentum: tempting
Agent floor / 4 voices, 1 veto
OrchestratorSESSION OPEN
Regime check passed. FTSE flat. Forwarding signal to the floor.
Context LeadCAUTION
No fresh RNS catalyst on BARC. This move is running on vibes, not news.
AnalystBORDERLINE
RSI 63.8 sits at the top of my band. Quality gate score: 4/10. I've seen this movie.
Risk AuditorVETO
Two cautions and no catalyst. Capital preserved. Try again tomorrow, human.
62 sessions. 0 impulsive trades let through.That would have been trade no. 63
Impulsive trades let through Sixty-two sessions and counting
04 / 04 HERITAGE

Stock-take, in Hindi.

My family's fabric business ran on paper and memory from 1988. Now warehouse workers count stock by speaking to their phones, fabric rolls in one hand. Tap the mic and watch.

UNI-NATION / VOICE STOCK ENTRYLIVE / MUMBAI + ULHASNAGAR
Transcript appears here
Stock register / parsed from speech
ProductLotPiecesMetres
Trovine Self174 thaan92.0
Samray Check092 taaka41.5
Why voice? Workers handle fabric rolls with both hands. Speech is the only free input. The parser understands thaan, taaka, metre and Hindi numbers, then fixes what speech-to-text gets wrong about fabric names.
~2,000 products / 650+ customers / every metre accounted for~$2m tracked
The schedule: projects under managementNote: all built solo
RefAssetDescriptionStatusCarrying value
01 OpsPilot The AI from chapter 2. Lives in Teams, reads 6,124+ documents, pulls live finance data, drafts answers I approve with one click. Production 97,000 lines
02 Trading Bot The four agents that just vetoed you. 62 sessions in: zero impulsive trades let through. That's the point. Paper trading 0 bad trades
03 Uni-Nation Digital Website, inventory system, and the Hindi voice stock entry from chapter 4, for the family fabric business running since 1988. Live ~$2m tracked
04 WhatsApp Smart 2.5 billion people use WhatsApp daily with zero productivity tools. A Chrome extension that adds scheduling, AI replies and summaries. Messages never leave the browser. Prototype 10 features built
05 SCS Cockpit The study engine I built to pass my own CIMA Strategic Case Study: timed drills, spaced repetition, zero dependencies. Sat the exam in May 2026. Retired, job done 4,200 lines / 0 deps
Total: five public systems, plus a drawer of internal tools built at work that stay off the booksAll in production use, or honest about not being
Experience: the day jobFinance and operations, one operator

Most of what's above is after-hours. By day I run finance and operations for a UK tech scale-up, on my own, across three countries. A few figures from the books.

3countries of finance and ops, run as the sole function
50+live dashboards on product, marketing and finance data
40+hrsof manual reporting automated away every month
65%cut from a core platform's cost by working the numbers
15to2page contract redesign: six-month stalls became one-week signings
7multi-currency accounts, GBP USD PKR
Supporting exhibitsClick to inspect
OpsPilot dashboard inside a laptop frame
OpsPilotExhibit A / AI operations
Trading bot terminal with signal dashboard
Trading BotExhibit B / Multi-agent AI
Uni-Nation website on desktop and phone
Uni-Nation DigitalExhibit C / Family business
WhatsApp Smart extension UI in a browser frame
WhatsApp SmartExhibit D / Chrome extension
Want the detail behind any of these?Correspondence

Happy to walk through how any of these were built. Email me, or find me on LinkedIn.

Note on the operatorBackground

Numbers by training. Systems by temperament.

I run finance and operations for a UK tech scale-up: one person covering accounts, contracts, HR, pricing and data across three countries. The software started because I got tired of waiting for tools that didn't exist. So I built them, and never really stopped.

  • N.1Finance and operations manager at a B2B tech company, sole finance function across UK, US and Pakistan entities since 2021.
  • N.2CIMA finalist: sat the Strategic Case Study in May 2026, alongside an MSc in Finance from Loughborough.
  • N.350+ live dashboards built on product, marketing and finance data; reporting automation that saves 40+ hours a month.
  • N.4Grew up in the family textile business in India. Commercial instinct came before the qualifications did.
  • N.5Builds in Python and TypeScript at night: AI agents, search pipelines, dashboards, the occasional Chrome extension.
  • N.6Four languages (English, Hindi, Sindhi, Marathi) and a school governor at a local primary. Taught himself the rest: SQL, Power BI, the AI stack.