Susanna Sargsyan
Verified Expert in Engineering
Machine Learning Engineer and Developer
Yerevan, Armenia
Toptal member since September 15, 2026
Susanna is a senior machine learning engineer with nearly 5 years of experience building and deploying scalable AI systems for clients, including Plat.AI and a robotics/autonomous systems company. Her core expertise spans advanced computer vision and generative AI, supported by robust MLOps engineering using Vertex AI, Kubeflow Pipelines, and GCP. Susanna thrives in fast-paced startup environments, where she recently architected a production ML platform supporting multiple prediction services.
Portfolio
Experience
- Machine Learning - 5 years
- PyTorch - 5 years
- Deep Learning - 5 years
- Python - 5 years
- OpenCV - 5 years
- Computer Vision - 4 years
- NVIDIA TensorRT - 4 years
- 3D Scene Reconstruction - 2 years
Preferred Environment
Vertex AI, Google Cloud Platform, Docker, Linux, Open Neural Network Exchange (ONNX), NVIDIA TensorRT, Git, PyTorch Lightning, OpenCV, Retrieval-augmented Generation (RAG)
The most amazing...
...ML platform I've architected supports multiple prediction services, including revenue forecasting and anomaly detection, using Vertex AI and Kubeflow.
Work Experience
Senior Machine Learning Engineer
Robotics and Autonomous Systems Company
- Engineered an end-to-end inertial navigation model that predicts autonomous system positioning directly from raw sensor data.
- Engineered a simulation-to-reality pipeline using NVIDIA Isaac Sim and ArduPilot, automating the capture of annotated image datasets with metadata for robotics simulation and synthetic data.
- Integrated a pipeline using SfM (COLMAP) for world model reconstruction and global retrieval for 6-DoF camera pose estimation via hierarchical localization.
- Developed 3D Gaussian Splatting pipelines for advanced scene reconstruction, leveraging photogrammetric outputs from Metashape to generate high-fidelity, geometrically precise 3D representations.
- Designed predictive machine learning models for precise battery consumption estimation across diverse trajectories, enabling robust, energy-aware mission planning.
- Engineered lightweight, high-performance image matching pipelines optimized for resource-constrained edge hardware to ensure robust navigation in GPS-denied environments.
- Developed deep learning models for feature-richness assessment to proactively identify optimal data regions and minimize failure rates in navigation and localization stacks.
- Engineered preprocessing workflows for aerial-to-map imagery alignment, significantly increasing matching success rates and overall navigation robustness in cross-domain data alignment.
- Designed and trained multi-object tracking models incorporating spatial constraints to enhance tracking stability in complex, dynamic environments for object tracking and re-identification.
- Researched and fine-tuned state-of-the-art generative models on internal proprietary datasets to improve perceptual quality for downstream computer vision tasks in image and video super-resolution.
Teaching Associate
American University of Armenia
- Held weekly office hours to mentor students through complex quantitative concepts, optimization techniques, and analytical problem-solving.
- Provided detailed technical feedback and academic support to students, improving overall comprehension of statistical and mathematical methods.
- Evaluated midterm and final examinations, ensuring consistent and rigorous grading standards across core mathematical modules.
Machine Learning Engineer
Plat.AI
- Designed and implemented end-to-end machine learning pipelines for loan default prediction, covering data preprocessing, feature engineering, exploratory analysis, model training, evaluation, and validation.
- Built predictive models using traditional machine learning algorithms and neural networks to support credit risk assessment and decision-making.
- Developed deep learning models for personalized insurance pricing based on customer behavior and risk profiles, ensuring scalability and alignment with business constraints.
Financial Analyst
Cognaize
- Validated extracted financial metrics against strict accounting rules to maintain high data accuracy across platform outputs.
- Evaluated complex financial data and loan covenants to support risk assessment and institutional client decision-making.
- Transformed unstructured financial documents into structured datasets to streamline automated client reporting workflows.
Experience
Agentic ML Platform & Gemini Router on Vertex AI
6-DoF Camera Pose Estimation & Spatial Metadata Extraction
AI Auto Part Cataloger & Listing Generator
https://youtu.be/fCMdIGvljYIEducation
Master's Degree in Statistics
Yerevan State University - Yerevan, Armenia
Bachelor's Degree in Mathematics
Yerevan State University - Yerevan, Armenia
Certifications
Ultimate AWS Certified Generative AI Developer Professional
Udemy
Skills
Libraries/APIs
PyTorch, PyTorch Lightning, Scikit-learn, XGBoost, OpenCV, NumPy, Pandas, SFML, Mypy, REST APIs
Tools
Open Neural Network Exchange (ONNX), Git, Pytest
Languages
Python
Frameworks
LightGBM, Streamlit, LlamaIndex, LangGraph
Storage
Data Pipelines
Paradigms
Anomaly Detection, Model Context Protocol (MCP)
Platforms
Vertex AI, Docker, Linux, Kubeflow, Google Cloud Platform (GCP), Amazon Web Services (AWS)
Other
Machine Learning, Deep Learning, Computer Vision, 3D Scene Reconstruction, Feature Engineering, Statistical Validation, Data Analysis, Exploratory Data Analysis, Data Engineering, Data Science, Model Evaluation, Model Development, Image Processing, AI Model Training, Transformers, NVIDIA Isaac Sim, COLMAP, NVIDIA TensorRT, Data Visualization, Google Cloud Platform, Agentic AI Systems, Risk Assessment, Credit Scoring, Gaussian Splatting, Model Monitoring, LangChain, LoRa, QLoRA, FAISS, Vector Search, Inference Optimization, Ruff, Linear Algebra, Probability Theory, Statistics, Bayesian Statistics, Multivariate Statistics, Calculus, Stochastic Modeling, Financial Math, Mentorship, Financial Analysis, Financial Reporting, Generative Artificial Intelligence (GenAI), Retrieval-augmented Generation (RAG), Large Language Models (LLMs), Gemini, FastAPI, Time Series, Optimization, Time Series Forecasting, Artificial Intelligence (AI), Multimodal GenAI, Hugging Face, Supervised Fine-tuning (SFT), Multimodal Models, APIs, Performance Optimization, ArduPilot, Model Registry, Model Deployment, Tool Calling, Machine Learning Operations (MLOps), eCommerce
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring