Bruno Saldivar
Portrait of Bruno Saldivar
Bruno SaldivarIndependent engineer · Berlin · Remote

Quantitative Developer · Agentic AI Engineer

I build trading systems, energy research platforms and AI agent platforms, and measure whether they work.

Four proprietary systems, each built end to end. Below, each one says what problem it solves and shows how it works.

The four systems

What each one solves, and how it works.

01Proprietary platform · live on MT5

Private Trading Research & Execution Platform

The problem
Most trading ideas are never tested against real costs, real risk limits and a real broker before money is on the line.
What it does
It takes a market idea from data to research, backtesting, risk checks and live orders on MT5, then checks the results against the broker's records.

Proprietary · built end to end

1,217order records from live execution on MT5, with profit and loss reconciled against the broker to the cent.

This describes the engineering, never the strategy.

Details and evidence →
How it works · 6 steps
How the trading platform works, in six steps Top to bottom. One, market data: ticks and candles from MT5, checked for gaps. Two, research: market hypotheses written as testable rules. Three, backtesting with realistic fills and broker costs. Four, risk: position sizing and drawdown limits. Five, live execution: orders sent to MT5, with order state that survives a restart. Six, reconciliation and diagnostics: profit and loss matched to the broker, to the cent. Expected versus actual results feed back into research. expected vs actual 1 Market data MT5 ticks and candles, gap checks 2 Research market hypotheses as testable rules 3 Backtesting realistic fills and broker costs 4 Risk position sizing, drawdown limits 5 Live execution orders on MT5, state survives restarts 6 Reconciliation & diagnostics P&L matched to the broker, to the cent
Simplified. What live trading shows, expected against actual, goes back into research.

02Proprietary product · voice execution on MT5

Trading Execution Copilot

Augmented intelligence for trading: the trader decides, the copilot scores and executes.

The problem
When a trader places orders by hand under time pressure, one wrong size or price can cost real money.
What it does
The trader says the order out loud. The copilot turns it into a ticket, scores the setup on executed trade history, checks it, sizes the risk, reads it back and sends it to MT5 only after an explicit yes, then reconciles it with the broker. The human stays in the loop for every order.

Proprietary · built end to end

Nothing reaches MT5 without the trader's explicit yes, and every order is read back first.

Details and evidence →
How it works · 7 steps
How the trading execution copilot works, in seven steps Top to bottom. One, voice order: the trader dictates the order and it becomes a ticket. Two, validation: the ticket is checked first. Three, risk sizing: the position size is set from the risk limit. Four, read back: the order is repeated back to the trader. Five, explicit confirmation: nothing moves without a clear yes; without it, nothing is sent. Six, the order is sent to MT5. Seven, broker reconciliation: the order counts as done once the broker confirms it. 1 Voice order dictated, turned into a ticket 2 Validation the ticket is checked first 3 Risk sizing size set from the risk limit 4 Read back repeated back to the trader 5 Explicit confirmation nothing moves without a clear yes no → not sent 6 Sent to MT5 only after the yes 7 Broker reconciliation done once the broker confirms
The trader decides the trade. The copilot makes sure the order that goes out is the one that was meant.

03Proprietary platform · deployed on-site in Oxford, UK (Sep - Oct 2026)

Energy Procurement Research & Backtesting Platform

The problem
I worked on-site in Oxford, UK, for a large energy buyer that spends heavily on power and gas every year and needed to lower its purchase cost.
What it does
A research and backtesting platform for power and gas procurement in European markets. It turns each buying idea into a testable rule, replays it on real exchange order-book and trade data with costs, and compares the result with the buyer's current practice and a market benchmark.

Proprietary · built end to end

2.5%below the buyer's current practice in 3 of 3 quarters (fees excluded), in its first exploratory backtest.

Built in under a month. First applied on-site in Oxford, UK (Sep - Oct 2026), to a large energy buyer's procurement.

Data and detailed results stay confidential.

Details and evidence →
How it works · 6 steps
How the energy procurement research platform works, in six steps Top to bottom. One, market data: real exchange order book and trades from European power and gas markets. Two, a buying question: what the buyer needs to decide. Three, a testable hypothesis, stated before anything is measured. Four, backtest and replay, with costs included and point-in-time data. Five, comparison against two references side by side: the buyer's current practice and a market benchmark. Six, decision: the evidence behind the buying decision. 1 Market data real exchange order book and trades 2 Buying question what the buyer needs to decide 3 Testable hypothesis stated before anything is measured 4 Backtest and replay costs included, point in time 5 Current practice the buyer's way today Market benchmark a market reference 6 Decision evidence behind the buying decision
Simplified. Every hypothesis is compared with both references, never only with itself.

04Proprietary platform · AI coding agents

Agentic Workflow Orchestration Platform (AWOP)

Agentic workflow orchestration: AI agents ship code, independent agents review it, tests decide what merges.

Multi-agent orchestration and coordination: agents share task ownership, handoffs and review rules. MCP (Model Context Protocol) connects assistants to the workflow; evals measure runs by model and role. Self-healing mechanisms handle stalled work and failed handoffs within explicit limits.

The problem
AI coding agents write code fast, but without structure nobody can say what changed, whether it was tested or who checked it.
What it does
It coordinates several AI agents in parallel, each on its own branch, with test gates on every merge and independent review on every autopilot merge. Human decisions and run records stay part of the workflow.

Proprietary · built end to end

341tasks merged across the four projects by AI agents, every one through the test gate.

It develops the trading decision engine and execution copilot today, and it built the energy platform.

AWOP coordination, recovery and evidence →
How it works · 7 steps
How the agent platform works, in seven steps Top to bottom. One, a task, scoped with acceptance criteria. Two, a planner prepares the work for the agents. Three, several writer agents work in parallel, each in its own workspace and git branch. Four, each change must pass a test gate that compares it with main; a change that makes the tests worse goes back to its agent for rework. Five, independent review: a different agent checks the exact change. Six, merge into main, only after the gate and the review. Seven, record and measurement: a run log and metrics by model and role. own branch own branch own branch 1 Task scoped, with acceptance criteria 2 Planner plans the work for the agents 3 Agent A writes code Agent B writes code Agent C writes code 4 tests ≥ main ✓ tests ≥ main ✓ worse → rework 5 Independent review a different agent checks the diff 6 Merge into main only after gate and review 7 Record and measurement run log, metrics by model and role
Simplified. Independent review is mandatory on every autopilot merge; on the other routes a person decides.

How I work

Clear scope, real evidence, stated limits.

  1. Fixed scope

    What is in, what is out and how the work will be accepted, agreed before it starts.

  2. Tests as evidence

    A change is done when the tests prove it, not when it runs once on my machine.

  3. Numbers with their sample

    Every figure I report says how many cases stand behind it.

  4. Limits up front

    What a system does not do is written down, not discovered later.

Bruno Saldivar

Hire me

Hiring a quantitative developer or an agentic AI engineer?

I am open to remote roles worldwide, and to project work through my agentic engineering services.

Based in
Berlin, Germany
Since
Independent quantitative research since February 2021
Looking for
Quantitative development · Trading systems engineering · Agentic AI engineering · Energy trading · Systematic trading · Remote
Languages
Spanish native · English professional · German A2.2, working toward B1
bru.daytrader@gmail.com LinkedIn Book a 30-min intro Full background