Senior Applied ML Engineer · Multimodal AI, 3D & Vision · London

Building perception systems for spatial and physical AI.

6+ years shipping 3D reconstruction, real-time vision and generative-AI systems to production — digital twins of large-scale urban environments, ADAS on embedded hardware, and LLM-assisted document intelligence for regulatory compliance.

Get in touch Download CV LinkedIn
Portrait of Alexey Solovyov

Selected work

Three chapters of building perception and generative systems for production.

Senior Applied Research Engineer

Elm Europe

London · Nov 2023 — Present · European R&D arm of a PIF-backed Saudi digital-services leader

3D reconstruction & digital twins of large-scale urban environments.

  • Established and led the company’s 3D and spatial-AI direction from scratch — technical strategy, architecture and delivery across an internal team of three and a five-person external research collaboration.
  • Built a reusable 3D reconstruction pipeline deployed across 5 urban districts (~10 km² each), fusing satellite, drone & vehicle imagery with hundreds of millions of LiDAR points per district.
  • Designed a district-scale 3DGS architecture with 90%+ cross-sensor coverage, pose refinement and real-time rendering, complemented by diffusion-based refinement for incomplete or dynamic regions.
  • Extended the digital twin into an LLM-assisted building-compliance product — grounded retrieval, structured extraction & tool-based validation — now live with initial Saudi clients.
  • 3DGS
  • NeRF
  • Photogrammetry
  • Diffusion
  • RAG
  • Multi-Sensor Fusion

Senior Computer Vision Engineer

Arrival

Moscow → London · Feb 2021 — May 2023 · EV technology company backed by Hyundai, UPS & BlackRock

Real-time perception on ultra-low-power edge hardware.

  • Rebuilt bus-CCTV passenger counting — 23× faster, +18% accuracy on embedded hardware.
  • Shipped a driver-monitoring system (ADAS) to safety-certification readiness.
  • Co-built the edge SDK for model optimisation & deployment.
  • Real-time CV
  • Edge / Embedded
  • ADAS
  • ONNX
  • Quantization
  • C++

Deep Learning & Generative AI

Gradient · Niias

Moscow · Jul 2019 — Feb 2021 · Consumer photo app with 50M+ users; NIIAS — MIPT computer vision R&D lab

Generative photo-editing for a 50M+ user app.

  • Shipped on-device features: hair recolour, background & object removal, facial filters.
  • Optimised segmentation & generative models for mobile latency and quality.
  • Prototyped GANs, VAEs and identity-preserving 3D-based editing.
  • GANs
  • VAEs
  • Segmentation
  • Mobile ML
  • Image-to-image

Skills

3D & Spatial AI
3DGS / NeRF · Photogrammetry (SfM / MVS) · Camera Calibration · Multi-Sensor Fusion · Georeferencing
Vision & Generative Models
Detection · Segmentation · Pose Estimation · Action Recognition · Diffusion Models · GANs / VAEs
ML Systems & Compute
PyTorch · Distributed Training (DDP / FSDP) · Slurm · GPU Cloud Platforms · Model Serving · MLflow
Edge ML & Deployment
Embedded Inference · ONNX · TFLite · Quantisation · Latency / Memory Optimisation
Applied LLM Systems
RAG · Vector Search · Tool Calling · Structured Outputs · vLLM · LLM Evaluation
Engineering & Infra
Python · C++ · Docker / Kubernetes · FastAPI · SQL · CI/CD

Education MSc & BSc in Computer Science, HSE University (Moscow) — a top-tier program created with Yandex. 2015–2021. Co-created and taught ML/DL courses for 200 students; olympiad winner (HSE CS & National Math, 2015 & 2019).

Let’s talk

Open to senior / staff roles in spatial AI, robotics & AV perception, multimodal AI, and generative systems.

alexeyvladimirovich38@gmail.com LinkedIn Download CV