Dubai, UAE · builder

Finance-trained,tech-built.

Investment banker by training, builder by instinct. I ship AI systems, research prediction markets, and run a fleet of autonomous agents.

See the workGet in touch
SCROLL

01 — Who I am

I sit at the intersection of finance, AI, and product.

I learned rigor executing ~$1.7bn of deals at Barclays, then started building the tools I wished existed: AI agents that automate the work, a research engine that trades prediction markets, products that reach real users.

I move fast, ship v1s with known gaps, and iterate in the open. Building is the thing I can't not do.

New York University Abu DhabiBA in Business, Computer Science MinorAdmitted at a ~3.6% acceptance rate · full-ride scholarship (~$305,000) · bp Scholarship, 1 of 125 globally

02 — Experience

Where I've built.

Aug 2024 — Present

Barclays Investment Bank

ICB Graduate Analyst · Dubai

International Corporate Banking coverage. I own a 28-client MENA book and, on my own initiative, shipped internal AI agents that no one asked for.

~40%
team prep-time cut by self-built AI agents
$33M
annual income across a 28-client book
~$1.7bn
live financing mandates executed end-to-end
Jun — Sep 2023

Google

Large Client Sales, Data Measurement · London

Built a quantitative model linking Media Mix Modelling activity to client revenue and presented the method to EMEA measurement specialists.

$70M
growth opportunity surfaced
40+
ran a GenAI productivity workshop
Dec 2023 — Apr 2024

Ziina

Growth & Data Analytics · Dubai · YC W21

Rebuilt the analytics stack at a YC-backed fintech, moving the team from ad-hoc requests onto self-serve reporting.

15+
SQL dashboards shipped across Growth, Marketing, Ops
Jul — Aug 2022

Amazon

Operations · Dubai

Drove bottleneck analysis across EMEA operations, turning root-cause findings into process fixes.

~20%
year-on-year cut in defects (DPMO)
$5M
projected annual savings from damage reduction

03 — Selected work

Things I've shipped.

01Autonomous research system

Prediction Markets Engine

2026 — Present

An autonomous, LLM-driven research engine for prediction markets. Claude agents read ~50 sources per market (vs ~5 for a human trader), price binary events into a typed forecast, then size confidence-tiered, fractional-Kelly entries on the divergence from live odds.

  • →Full Python / SQLite stack across 4 strategy pillars: event synthesis, cross-market arbitrage, crypto, funding-rate capture
  • →Backtest-derived entry filters and out-of-sample resolution gates that retire any strategy without measurable edge
  • →Rigor over results — paper-traded throughout, validate before any capital
Claude agentsPythonSQLitePolymarketBacktesting
02Internal tooling, self-initiated

AI Agents at Barclays

2024 — Present

Internal AI agents the desk never asked for. LLM workflows scaffolded over the bank's proprietary data across five processes, with human-in-the-loop review and a feedback log that tightens prompts over time.

  • →5 workflows: KYC, client research, cross-border deal prep, legal review, portfolio committee drafts
  • →~40% cut in team prep time
  • →Wrote the rollout playbook now run by adjacent desks
LLM scaffoldingRetrievalPrompt evaluationProprietary data
03Co-founded consumer app

Thunder

2023 — 2024

A student events app to help people meet. Co-founded and led discovery-to-MVP, iterating product scope on real user feedback.

  • →400+ early-access sign-ups
  • →Top 10 of 380 teams at SkillUp Compete (MIT Enterprise Forum Pan Arab & UAE Ministry of Economy)
Product0 to 1User researchGTM

04 — The fleet

I don't just use AI. I run a fleet of it.

Beyond the day job, I run a personal fleet of autonomous AI agents. They research markets, generate content, build products, and ship work end-to-end.

Quantresearches & paper-trades prediction markets
Studiowrites, designs, and renders creative work
Builderscaffolds and ships product surfaces
Operatorruns the back office on autopilot

This very site was scoped, designed, and deployed autonomously by one of them.

05 — Toolkit

How I build.

AI Builder

  • Claude Code
  • Cursor
  • Python
  • Multi-agent LLM systems
  • Web search & retrieval / RAG
  • Typed / structured outputs
  • Prompt evaluation
  • Ships end-to-end demos (0 to 1)

AI Deployment

  • Turning ambiguous problems into shipped AI workflows
  • Building over proprietary data
  • Human-in-the-loop review
  • Feedback-logged iteration
  • Adoption across teams

Data & Product

  • SQL
  • R
  • Looker · Mode
  • Experimentation & measurement
  • Cohort & funnel analysis
  • Financial modelling (DCF / LBO)
  • Product discovery & roadmapping

06 — Let's build

Have something
worth building?

I'm always up for a conversation about AI, markets, product, or anything ambitious. The fastest way to reach me is email.

akshattotla@gmail.comLinkedIn ↗