Soham Ranade

AI Researcher

Palo Alto, California, United States5 yrs 11 mos experience

Most Likely To SwitchHighly Stable

Key Highlights

Led development of a GenAI-powered finance chat platform.
Architected secure user management for thousands of users.
Expert in fine-tuning large language models for enterprise solutions.

Stackforce AI infers this person is a Fintech-focused Machine Learning Engineer with strong product management capabilities.

Contact

ssr2182@columbia.edu LinkedIn

Skills

Core Skills

Machine LearningNatural Language Processing (nlp)

Other Skills

DockerFastAPIFine TuningLarge Language Models (LLM)Python (Programming Language)SQL

About

I'm a Machine Learning Engineer and Technical Product Owner passionate about building intelligent systems that bridge the gap between cutting-edge AI and real-world business impact. At Vianai Systems, I lead the development of Hila, a GenAI-powered finance chat platform that's transforming how enterprises interact with data—securing multi-million dollar contracts along the way. With a background spanning query response optimization, natural language disambiguation and agentic orchestration, I design systems that not only perform, but also earn user trust. My work includes architecting secure user management for thousands of concurrent users, creating metrics to rigorously evaluate LLM output quality, and deploying robust forecasting models to support critical financial decisions. I thrive at the intersection of product thinking and technical execution—partnering with engineers, designers, and business stakeholders to bring bold ideas to life, from notebook to production. Whether it's fine-tuning large language models or optimizing Kubernetes-backed ML pipelines, I bring a hands-on, systems-level mindset to everything I do.

Experience

Vianai systems, inc.

3 roles

Machine Learning Engineer

Promoted

Dec 2023 – Present · 2 yrs 3 mos

Collaborated with Product Management and Engineering to productionize new features for Conversational Finance, including in-context conversations, dynamic charting, and disambiguation. The application acts as a chat interface atop enterprise data, helping visualize financial insights for institutions.
Fine-tuned a LLM to assist with answer verification and completeness, identifying key performance metrics and developing a finetune model for verification.
Experimented with new LLM models and created benchmarks and test modules for rapid performance feedback.
Productionized the integration of the Structured Q&A feature by ensuring the modularity of subprocesses and connecting them through an orchestration pipeline.
Developed backend services using FastAPI and Docker to enable drift detection, validation, explainability, and performance monitoring at scale on the VIANOPS platform, which handles approximately 1 billion rows of data daily and is used by several Fortune 500 companies.
Architected and implemented a secure backend login workflow for the LLM-based financial research tool Hila, supporting a user base of approximately 3,500 users.
Led Agile development as Scrum Master, facilitating sprint planning, daily stand-ups, and retrospectives for the development team.

Large Language Models (LLM)Fine TuningMachine LearningNatural Language Processing (NLP)Python (Programming Language)SQL

Data Scientist

Jun 2022 – Present · 3 yrs 9 mos

Created various demo modules with the product to showcase different features across releases and help with customer onboarding from fortune 500 companies.
Contributed to the development of ModelFactory framework for easier management of Vianai supported models within the platform.
Designed and maintained pre- and post-processors with pandas, scikit-learn, and Spark frameworks to facilitate smooth data transformation across complex models on the no-code Vianai platform.
Experiment with different language models like BERT, SBERT and created pipelines for question answering, summarization, etc.
Implemented Stats-based, ARIMA, and Prophet model training for time series modeling on the platform.
Assisted in developing test plans for UAT testing by creating mock datasets and deploying inference pipelines.

Python (Programming Language)SQLMachine Learning

Data Science Intern

Jun 2021 – May 2022 · 11 mos

Built an end to end uncertainty based classification module for the company's Machine Learning platform.
Created an end to end hierarchical zero shot classifier module to assist in text classification and training.
Trained the existing zero shot model with only 100 examples to predict the classes with 66% accuracy.

Python (Programming Language)SQL

Columbia university in the city of new york

2 roles

Research Assistant

Jan 2022 – May 2022 · 4 mos

* Working in the Robotics and Rehabilitation Robotics Lab to create a passive multipoint crutch device for increased support during the crutch walking gait.

Course Assistant for Intro to Robotics

Sep 2021 – May 2022 · 8 mos

Created and Developed Assignment modules for the course.
Assisted, guided and graded the students for successful completion of the course and ensured a clear communication between the Professor and the students.

Amazon

2 roles

Area Manager

Jul 2020 – Dec 2020 · 5 mos

Managed ~300 warehouse associates across shifts to ensure smooth processing of customer orders.
Oversaw fulfilment process; identified opportunities to mitigate barriers and improve operational efficiency.
Acquired and analysed data using SQL, MACROS to help cross functional teams in improving various processes.

SQL

Area Manager Intern

Jan 2020 – Jun 2020 · 5 mos

Developed and implemented an inventory placement strategy which increased the warehouse savings by 1.3%.
Simplified various processes by building MACROS and writing SQL queries.

Indian institute of technology, bombay

Research Intern

May 2019 – Jul 2019 · 2 mos · Mumbai, Maharashtra, India

Designed a test rig for a gas turbine seal to withstand high temperature-pressure for GE Aerospace.
Fabricated and simulated prototype to withstand 1000 °C, 7MPa conditions using SOLIDWORKS, ANSYS; results showed an estimated increase in efficiency to 7% with respect to the current turbine models.

Tata technologies

Summer Intern

May 2018 – Jul 2018 · 2 mos · Pune Area, India

Designed a modular fuse box to facilitate the maintenance of TATA Bolt and Zest models
Modelled the prototype using Catia to reduce maintenance costs.
Built a license plate recognition software for Indian License Plates using OpenCV, Python, and
TesseractOCR.

Team robocon bits pilani

Team Captain

Apr 2018 – Apr 2019 · 1 yr

Designed two robots (one manual and one autonomous) as part of a pan India robotics competition.
Headed the electronics department of 10 people and designed PCBs for all sensor-actuator integration.
Captained the team in pre-final year and conceived research projects to initiate student-teacher collaboration.