Who we are

SIMYL Research

SIMYL Research is a software engineering consultancy specializing in high-performance backends and APIs for systems where milliseconds matter and precision is essential, with a focus on sports betting, fintech, and B2B SaaS. We also build quantitative models and simulations when the problem lies in the computation.

SIMYL RESEARCH S.A.S., incorporated in Colombia in 2025.

Why SIMYL exists

SIMYL Research exists to build and improve software with the rigor we learned in academia and in industry: measure before giving an opinion, agree in writing on what is delivered, and publish how we check it. At the helm is a leadership team that combines doctoral training, university teaching, and years of systems in production, plus more engineers with whom we build each engagement; we prefer few engagements with one owner from start to finish. We look for clients who want to understand their system, not just delegate it: people who review our benchmarks, read our open-source code, and ask us for evidence. If we don’t master something, we say so on the first call.

Our approach

Our focus is the backend: architecture, business logic, performance, and integrations. We can deliver complete applications, including the web interface, when the core of the problem lies in business logic and data. If the project requires product design or specialized front-end work, we say so in the proposal and define who takes it on.

We work by three criteria:

Speed. We deliver tangible results in short cycles. We don’t sell endless projects.

Impact. We attack the real problem, not just the symptoms. We prioritize what creates immediate business value.

Quality. Performance from the start. Clean, tested, maintainable code. We don’t build throwaway prototypes; we do build production-ready first versions with a bounded scope and written acceptance criteria.

Experience is not what happens to a man; it is what a man does with what happens to him.

Aldous Huxley, Texts and Pretexts: An Anthology with Commentaries (1932), introduction. Source.
Experience becomes useful when we examine our decisions and share what we learned. That is why we explain both the results and the limits of our work.

Our process

  1. Understand. A free 30-minute discovery call to learn about the problem, the system, and the constraints.
  2. Agree in writing. Scope, deliverables, acceptance criteria, and payment terms before we start. The models are described in How we work.
  3. Build or intervene. With one technical owner from start to finish, frequent deliveries, and review of all the code we deliver.
  4. Validate with measurements. We measure before and after every agreed intervention, with the method and conditions stated.
  5. Hand over. Documentation, sessions with your team, and a stabilization period included in development projects.

Team experience

Members of our team have prior experience in:

  • Algorithms and data structures: we review them together with queries, concurrency, and infrastructure; the measurements determine what to change first.
  • Distributed systems: architectures that scale without collapsing, designed for real load rather than demos.
  • Sports betting and fintech: betting-market calculations and financial applications where milliseconds matter.
  • Real-time systems: low latency, event processing, reactive architectures.
  • Modeling and simulation: Monte Carlo, machine learning, mathematical optimization.
  • Oil and gas: production simulation from the wellbore to the separator, and a domain-specific language for expressing reservoir correlations (PVT and fluid properties), with heavy numerical computing.
  • Combinatorial optimization: graph problems, network flows, scheduling, resource allocation.
  • Modern infrastructure: Kubernetes, observability, CI/CD.

Team track record

SIMYL Research was incorporated in 2025. These works were designed or led by members of the team that now runs it, before the company or outside it; we do not name companies or clients, for confidentiality. The ones with their own entry are in Experience.

  • Trading (2024–2026): Monte Carlo simulation engines optimized to cut execution time.
  • Sports betting (2020–2023): a cache protocol for a real-time API, in production since 2020 and validated up to 100,000 requests per second in 2023 tests (entry).
  • Fintech (2017): p99 latency cut by more than 40% in production on critical APIs.
  • Property (2017): multi-tenant backends for a property and condominium management platform (entry).
  • Migration (2017): from a Yardi-based system to essensys without stopping the business or losing data (entry).
  • Energy (2016–2018): well-to-separator production simulation and a domain-specific language with 188 PVT correlations, published as open source (entry).
  • Remote-work monitoring (2014–2015): collector client and collector server in C++ (entry).

Who do we work with?

We primarily work with:

  • CTOs, technical leads, and engineering or data teams at companies already in production
  • Tech startups that need to build robust products from the start
  • Established companies facing complex technical problems or needing to modernize legacy systems

Technologies

We primarily work with:

  • Very high-performance microservices: modern C++ (17/20/23) as the primary language, Go, Rust
  • Web servers: Go, Ruby on Rails
  • Mobile apps: Flutter, React Native
  • Data analysis: R (tidyverse, Shiny), Python for automation and analysis
  • Infrastructure: Kubernetes, Docker, cloud providers (AWS, GCP, Azure)
  • Databases: PostgreSQL, Redis, ClickHouse, time-series databases
  • Observability: Prometheus, Grafana, Jaeger, ELK

Open source

What we publish can be read, run, and criticized:

  • cpp-cache: concurrent in-memory cache for C++20 processes, with a Promela/SPIN formal model in the repository; 2026 measurements in a declared environment in Benchmarks.
  • gateway_cache: the Go implementation of the same protocol, published in 2020 and in production since then; its version 3.1.0 is the one measured in Benchmarks.
  • Aleph-w: data structures and algorithms library in C++.
  • go-breaker: circuit breaker for Go services that trips on memory and latency thresholds, with trend analysis and OpsGenie alerts.
  • uconv: units and physical-quantities library in C++20, with strongly typed quantities at compile time and a DSL to define units and conversions.
  • back-pvt: PVT characterization of fluids in C++, physical properties as a function of pressure and temperature.

Why SIMYL?

We are not a traditional development agency. We are technical specialists who:

  • measure before giving an opinion and show the evidence;
  • write production code, not throwaway prototypes;
  • prioritize quality and long-term maintainability;
  • communicate technical decisions and their business consequences clearly;
  • if we don’t master something, we tell you: we would rather say no than deliver something mediocre.

Meet the team

The team page shows who we are, what each of us is responsible for, and where we come from.

See the team

Request the discovery call and tell us what you need to solve.