Software engineering · measured and published

We build and improve the software that powers your business

We help you turn a need in your business into a tool you can use, on the web or on the phone, and improve the systems you already have. We start by understanding how you work and what you want to achieve; we measure before and after every change and publish how we do it.

Where to start

Two entry points with a published scope and reference price. Pick the one that describes your situation.

For founders and product leads

Create a tool for my business

Are there tasks that eat your time, data scattered across spreadsheets, or an idea you want to get off the ground? We help you clarify what you need and build a first useful version, on the web or as a phone app, to test it with the people who will use it. You don't need to arrive with a solution already defined.

  • About six weeks: two of definition and four of build, with a demo every week.
  • Fixed price by milestones; as a reference, USD 8,000 for a complete first web version.
  • The code is yours, documented and ready for any team to continue.

Get a quote for a first version in 6 weeks

Reply within 1–2 business days · fixed price by milestones · direct contact with senior engineers

For CTOs and technical leads

Improve a system I already use

We fix latency bottlenecks, failing APIs, and cloud costs that grow faster than the business, and we take quantitative models to production, with C++ (C++17/20/23), Go, Rust, and R. We specialize in the performance of backends and APIs for sports betting, fintech, and B2B SaaS.

  • Performance and architecture diagnostic in two weeks; as a reference, USD 2,900.
  • A 4-hour consulting session for one specific question; as a reference, USD 600.
  • Read-only access, NDA, and a report your team can execute without us.

Request the diagnostic (USD 2,900) Consulting session (USD 600)

We measure before and after every change, under the same conditions: same load, same environment, same tools.

  • Since 2020

    in production: the open-source cache protocol we explain on the blog.

  • 3.7 M requests/s

    cpp-cache with one worker, serving from the cache; benchmark of 18 October 2026. See the data

  • 1–2 business days

    response time for new inquiries. Working remotely across the Americas and Europe.

How we collaborate

  1. A free 30-minute call to understand your case.
  2. A written proposal with fixed scope, timeline, and price.
  3. Work by milestones, with deliveries you can test.
  4. Delivery with documentation and handover; the code is yours.

See how we work

Check it yourself

Three public pieces to verify what we say before you write to us: dated results on a declared machine, code you can run, and the method we measure with.

Reproducible benchmark

runs
42
errors
0
declared machine
Oct 2026

Two caches measured under the same protocol. Raw data and measurement code published.

See the results

Open source

Six projects you can read, run, and criticize: two caches with the same protocol, data structures, a circuit breaker, physical units, and PVT correlations.

See the six projects

How we measure

Three kinds of figures (benchmark, production, estimate), a before-and-after protocol, and downloadable templates for the measurement report and the acceptance criteria.

Read the method

Services

Three lines focused on high-performance critical systems and applied modeling. Engage us for a diagnostic, a milestone-based project, or monthly advisory.

Technical consulting for performance and architecture

  • Diagnostic of APIs and critical systems
  • Architecture and scalability improvement plans
  • Gradual legacy modernization, minimizing disruption

See the diagnostic

High-performance software development

  • Backends and APIs; web applications and apps for iPhone and Android
  • Standalone programs and command-line tools in C++ or Rust
  • Data processing and integration with existing systems

See project types

Modeling and analysis

  • Monte Carlo simulations and risk analysis
  • Forecasting and optimization models
  • Dashboards and decision automation

See the modeling service

Technical capabilities:
API backends & microservices Backend architecture Elastic distributed systems Sports betting, fintech, and B2B SaaS Cloud / Kubernetes Databases Data science & R (tidyverse) Machine learning

Technologies we regularly use in our backends and systems:

Team track record

SIMYL Research was incorporated in 2025. This earlier work was led by members of the team that now runs it and shows how we approach a problem; we do not name companies or clients.

Cache and architecture for a sports betting API

2023 tests · protocol and architecture design: the backend of a real-time sports betting API, optimized by combining high-performance in-memory caching with hash-based routing across multiple gateways, so that most repeated requests were served from cache without recomputing prices. Under controlled test conditions in 2023, this architecture demonstrated a capacity of up to 100,000 requests per second with p99 latency below 500 ms.

Multi-tenant backend for property and condominium management

2017 · design and implementation: backend APIs for a multi-tenant property and condo management platform, handling contracts, payments, notifications, and reporting, built to grow gradually without service interruptions.

Migration of a Yardi-based system to essensys without service interruption

2017 · responsible for the migration: an operation running on Yardi had to move to essensys without stopping the service or losing data; the data was extracted through Yardi's integration interface and the service kept running during the transition. Migration completed without interruption or data loss.

Well-to-separator production simulation and a DSL for PVT correlations

2016–2018 · lead developer at a startup: a backend for simulating production from the bottom of the well to the separator and a domain-specific language to define PVT correlations with checked units and ranges. 188 correlations in 35 families published as free software in back-pvt; the units library uconv was born there.

Graph-based analyzer of productive chains

2012–2016 · lead of the analyzer, as an external collaborator of a public research center for free technologies: a graph of productive chains, from raw materials to final goods, built from industrial-registry data to locate their weak points, such as cut points and minimum cuts by product line. C++ code on Aleph-w, published as free software.

See all entries

Open source

What we publish can be read, run, and criticized. Six projects, by name, purpose, and status.

API performance and resilience

Related service: Technical consulting for performance and architecture

gateway_cache

Collapses simultaneous requests for the same data into a single call to the origin, so a traffic spike does not overwhelm the database.

The Go implementation of the same cache protocol, designed for API gateways.

Go · MIT license · published in 2020 · in production since then on the sports data platform of Genius Sports

cpp-cache

The same protocol inside a C++ process: one call to the origin per key, however many requests ask for it at once.

Concurrent in-memory cache for C++20 processes, with a single-flight protocol against request stampedes.

C++20 · MIT license · Promela/SPIN formal model included in the repository · 2026 benchmarks published

go-breaker

Cuts traffic to a degrading service in time, so it does not drag down the rest of the system.

Circuit breaker for Go services: it trips on memory and latency thresholds, with trend analysis and OpsGenie alerts.

Go · MIT license · tests included

The 2023 performance figures correspond to the Go implementation; we do not extrapolate them to cpp-cache. The 2026 measurements of both libraries, under the same protocol on a declared machine, are in Benchmarks.

Scientific computing and modeling

Related service: Modeling and analysis

Aleph-w

Algorithms and data structures for graph, network, and optimization problems, such as the analysis of productive chains.

Data structures and algorithms library in C++ with an emphasis on functional programming.

C++ · MIT license · under active development

uconv

Prevents unit errors in engineering calculations: the compiler rejects, for example, adding a pressure to a temperature.

Units and physical-quantities library in C++20, with strongly typed quantities at compile time and a DSL to define units and conversions.

C++20 · GPL-3.0 license · continuous integration

back-pvt

Computes the properties of reservoir fluids from pressure and temperature.

PVT characterization of fluids: physical properties as a function of pressure and temperature, with its test suite.

C++ · GPL-3.0 license · tests included

Licensing: we do not include GPL-licensed code in your product unless we agree to it in writing.

Who we are

High-performance software engineering

At SIMYL Research, we combine academic rigor with industrial experience to build critical systems where every millisecond and every cent counts.

  • Efficient code: we optimize algorithms to reduce your cloud costs, and measure before and after.
  • No intermediaries: a small senior engineering team, with a fixed scope and price.
  • Transparency: milestone-based payments with written acceptance criteria.
  • Precision: statistical and machine-learning models to quantify uncertainty and decide under explicit assumptions.

Want to know who you will work with?

Meet the team More about who we are

Frequently asked questions
I don't know exactly what I need. Can I still write to you?

Yes. Tell us what you do today and what is hard; on the 30-minute call we work out together whether it is worth building something, using an existing tool, or changing the process. If we are not the best option, we tell you right there.

What happens on the 30-minute call?

We listen to your case, ask questions, and tell you whether we can help and how. It is free and without commitment; afterwards you receive in writing what we propose, with scope, timeline, and price.

Do you also build the web interface and mobile apps?

Yes, of course. We build the web interface and the iPhone and Android apps, with a single codebase for both platforms, and publish them to the stores with your accounts. We have mobile engineers who work under our technical direction, with Flutter by default or React Native if your team already uses it. And we take special care of what you don’t see: the screen attracts the first users, but what decides whether they stay is that everything responds quickly, loses no data, and holds up as the business grows. In the agile MVP, the app is a separate module: as a reference, USD 5,000 and about three more weeks; the exact price depends on its complexity and is fixed in the definition phase. We also build it for a system you already have, as a milestone project.

Do you work hourly or per project?

It depends on how much uncertainty the project has. When the scope is stable, we work on a fixed price by milestones, with an advance payment and written acceptance criteria for each delivery. When uncertainty is high (integrations still to be discovered, requirements to validate), we work time-based with a budget cap and regular reviews, because fixing a price too early usually hurts both sides. For specific questions we offer a 4-hour session for USD 600 as a reference. The standard terms for each model are in How we work, and we adjust them in every proposal.

Do you cover post-delivery support?

Yes. Development projects include a stabilization period defined in the proposal and knowledge transfer to your team. After that, you can hire monthly advisory or a maintenance hour bundle; if you prefer an in-house team, we help you train or find the people who will maintain the system.

How do you invoice clients in Colombia and abroad?

We issue electronic invoices. Clients in Colombia are invoiced in Colombian pesos, with the amount fixed in the proposal; clients in other countries in US dollars or euros, paid by international bank transfer. Before signing, we tell you how taxes and withholdings apply in your country.

What is the minimum project size you work with?

We prioritize projects where we can make a clear impact on critical systems; as a reference, initiatives lasting from several weeks to several months. For smaller needs we offer a 4-hour consulting session and the performance and architecture diagnostic, with a fixed scope and a reference price. We take on few projects at a time so each one has a senior owner from start to finish.

Do you work only remotely, or also on-site?

We work remotely, on Bogotá time (UTC−5). For key milestones such as workshops, kick-offs, or audits, we can travel when the scope and budget justify it.

Can you work with the technologies we already use if they are not C++, Go or Rust?

Yes, as long as we have real experience with them. If we don’t master them, we’ll say so and focus on architecture, performance, and best practices, or help you define a migration path.

How do you handle confidentiality of our information and code?

We sign NDAs when needed, use repositories with restricted access, and follow security best practices. We do not reuse client-specific code across projects, and we publish nothing about your project unless the contract allows it or you authorize it.

Which languages do you work in?

We work in Spanish and English. Written deliverables are usually in English or Spanish, depending on what your team needs.

Who leads SIMYL Research

Every engagement has a named lead who designs it, runs it, and answers for the result.

Julian Segura

Co-founder · product leadership and agile methods

Turns business goals into fixed scopes and punctual iterative deliveries: defines the proposals, negotiates the terms, and follows every milestone with the client.

Leandro Rabindranath León

Co-founder · principal architect

Author of the open-source library Aleph-w and of the concurrent cache protocol in production since 2020; architecture in C++, Go, and Rust, after leading distributed-systems teams at Genius Sports.

Alex Drums

CTO · technical leadership and engineering

Senior backend engineer, specialist in Django and in the fast development of secure web platforms; leads the build of every first version.

See the team page

Courses and workshops

Hands-on training for teams already in production. Remote or on-site; we issue certificates of participation.

Modern C++ (C++17/20/23)

Smart pointers, move semantics, templates, and concurrency. 4–5 days.

High-performance computing

CPU caches, profiling, lock-free structures, and SIMD. 3 days.

Kubernetes for developers

Deployments, resources, autoscaling, and production troubleshooting. 3 days.

Tell us what your team wants to learn See the courses

Two typical situations, start to finish

Problem, intervention with scope and price, and a measurable result: this is how we approach an engagement, whether improving a system in production or creating a new one.

For CTOs · earlier work by the team, 2023 tests

Sports betting and real-time data

Problem
A real-time betting API received many repeated requests that recomputed prices at the origin; under traffic peaks, the stampede multiplied load and latency.
Intervention
Protocol and architecture design: a concurrent in-memory cache with a single-flight protocol and hash-based routing across several gateways, so repeated requests were served from cache without recomputing. Today that work starts with the performance and architecture diagnostic (USD 2,900), and the improvements are implemented in C++ or Go.
Result
Up to 100,000 requests per second with p99 below 500 ms in controlled tests in 2023, and more than 95% of hits in the first cache layer. The protocol, published as open source in 2020, is still in production.

See the full case

For founders · the published scope of the agile MVP

Fintech or B2B SaaS platform: a first version in six weeks

Problem
Launch a working product in six weeks on a fixed budget, without mortgaging future scalability or rewriting the platform as it grows.
Intervention
Agile management with weekly milestones; a web application in Django or Ruby on Rails to reach the market fast; the critical processing isolated from the design as a Go service, so it can scale without touching the rest.
Result
A first version in use in about six weeks, at a fixed price by milestones (as a reference, USD 8,000) and with the code in the client's name. This is the published scope of the agile MVP, not a delivered project.

See the full scope

Contact

Tell us what you need to solve; a short description is enough to start. We reply within 1–2 business days.

You can tell us what you do today, what is hard, or what you would like to be able to do. A short description is enough to start; you don't need to know which technology it takes.

We use your details to answer your inquiry and, if you ask, to prepare a proposal and follow up; submissions are processed by Formspree. Do not include credentials or confidential information. Details in the privacy notice.

Email: contact@simylresearch.com