Software Developer
CurrentEngineer on MYFRT, a digital-forensics platform investigators use to acquire, process and analyse mobile evidence. I work across the stack — from parsing forensic extractions and running AI/ML models over the evidence, to the APIs, cloud deployment and the interfaces examiners investigate through.
- Built the evidence-processing pipeline for mobile forensic data: ingesting UFED/UFDR (Cellebrite) extractions and E01 disk images, detecting format and version, parsing defensively, normalising into a common evidence model and persisting every record with provenance back to its source artefact.
- Integrated AI/ML models into the platform's processing workflows — YOLO-based object detection and facial recognition over extracted images and video for automated media triage.
- Ran local LLMs inside the application rather than calling hosted APIs, so sensitive case data never leaves the deployment; owned the inference integration, prompt and response handling, and the debugging when model output did not fit the workflow.
- Built live mobile device analysis: connecting to a handset and streaming its contents as the examination happens, with batched, virtualised rendering so an investigator can work through a dataset that is still growing.
- Designed the investigative review surfaces on top of the processing layer — search, filtering and case workflows modelled on how examiners actually work, so extracted artefacts are reviewable rather than a wall of raw output.
- Profiled and optimised the slow paths across query, payload and render on evidence sets large enough that naive approaches stop being viable; the same work applies to the model-inference queue.
- Designed and documented the REST APIs carrying data between acquisition, processing, analysis and reporting; kept modules decoupled with SOLID and low-level design patterns so a new extraction format or model is a new module, not a rewrite.
- Own production: investigating and fixing issues reported from live deployments, from a parser choking on an unfamiliar extraction to a model integration misbehaving on real data. Also built the admin panel and role-based access for teams handling sensitive case data.
- Contributed to the company's blockchain products, including transaction-monitoring work for crypto fraud and AML-style investigation workflows.
Domains owned
Mobile & digital forensicsUFED / E01 evidence processingAI/ML model integrationLocal LLM inferenceYOLO object detection & facial recognitionData parsing & pipelinesLive device analysisProduction debuggingAPI design
PythonFlaskNext.jsReactTypeScriptMongoDBREST APICloud deployment
