Acting as a Fractional CTO and Lead Engineer, delivering production-grade solutions in Quantitative Finance and Multimodal Generative AI pipelines.
š š¤šØšš”š§šš§šš§šš©š ššš”šš”šš š¦š¬š¦š§šš (š£šššµš¼š»/šš»š“š¶š»š²š²šæš¶š»š“)
ā¢Engineered a standalone šš¼š¶š»šš²š“šæš®šš¶š¼š» šš»š®š¹ššš¶š šš»š“š¶š»š² for pairs trading strategies, focusing on robustness and production deployment rather than ad-hoc scripts.
ā¢Implemented rigorous statistical tests (ADF, CADF) within a modular Python codebase for high-performance backtesting and signal generation.
ā¢Tech Stack: Python, NumPy, Statsmodels, Time-Series Analysis.
š š¦ššš-šš¢š¦š§šš ššš”-šš šš”šš„šš¦š§š„šØšš§šØš„š & ššØš§š¢š šš§šš¢š”
ā¢Architected a š°š¼šš-š²š³š³š¶š°š¶š²š»š, š½šæš¶šš®šš² š°š¹š¼šš± šš š²š»š“š¶š»š² for a content production studio. Migrated workflows from expensive public APIs to self-hosted, optimized open-source models, ensuring 100% data sovereignty and reducing operational costs.
ā¢š£šæš¶šš®šš² ššš šš»š³š²šæš²š»š°š² š¦š²šæšš¶š°š²: Deployed optimized open-weights models (Qwen-3) on GPU instances as a secure alternative to GPT-4. Engineered a ššššš¼šŗ šš¼š»šš²š
š š¢šæš°šµš²šššæš®šš¶š¼š» šš£š to drive automated script generation and roleplay simulation.
ā¢šš²š®š±š¹š²šš šš¼šŗš³ššØš šš®š°šøš²š»š±: Engineered a scalable image generation service by wrapping complex ComfyUI workflows into š„šš¦š§ šš£šš. This enabled the remote production team to trigger consistent character generation programmatically without managing node graphs.
ā¢šššš¼šŗš®šš²š± šššš²š š£š¶š½š²š¹š¶š»š²: Integrated šš¼ššš©š¼š¶š°š² for high-fidelity TTS and utilized š¢š»š²š§šæš®š¶š»š²šæ to manage remote LoRA fine-tuning jobs, ensuring strict IP consistency across distributed workflows.
ā¢Tech Stack: Cloud GPU Inference (RunPod/AWS), Headless ComfyUI, API Design, Docker, Python, vLLM/Ollama.
