Machine Learning Engineer

FRND; Bengaluru; On-site; INR 2,800,000 - 4,600,000; Full time

Job summary

FRND is seeking a Machine Learning Engineer to design, train, and deploy scalable ML models across image, audio, and text modalities. You’ll own the full model lifecycle, from data curation to production inference, and collaborate with Backend, Product, and Moderation teams to address real‑world safety challenges.

You’ll work with Python, PyTorch or TensorFlow, and manage low‑latency services with Docker and Kubernetes, across multilingual data and multilingual geographies.

Qualifications

• 4–6 years of experience shipping ML models to production.
• Strong Python experience with PyTorch or TensorFlow.
• Deep expertise in at least one domain: CV, Audio/Speech, or NLP.
• Experience with transformer architectures and pretrained backbones.
• Proven production experience: serving, latency, throughput, monitoring.
• Familiarity with data at scale and ML eval/experimentation.

Responsibilities

• Design, train, and deploy production ML models across image/video, audio, and text.
• Own end-to-end model lifecycle from dataset design to retraining.
• Build low-latency inference services for real-time moderation.
• Define metrics for safety, language/geography coverage, and cost.

Skills

Python

PyTorch

TensorFlow

Docker

Kubernetes

SQL

PostgreSQL

Kafka

ClickHouse

AWS

GCP

Computer Vision

NLP

Multilingual data

Trust & Safety

Tools

PyTorch

TensorFlow

Docker

Kubernetes

Kafka

ClickHouse

PostgreSQL

AWS

GCP

Job description

At FRND, you’ll build, ship, and own the ML systems that help keep millions of real-time interactions safe, while also pushing the boundaries with new AI experiments.

FRND is built around live social interaction. People connect through 1:1 video calls, audio rooms, many-to-many conversations, and messaging across dozens of languages, all in real time.

As a Machine Learning Engineer, you’ll work on problems that are both high-scale and high-impact, taking ML problems from dataset design all the way to production inference. You’ll work closely with Backend, Product, and Moderatio teams to turn complex safety challenges into models that actually work in the real world.

What You’ll Do

• Design, train, and deploy production ML models across image/video, audio, and text.
• Own the complete model lifecycle, from dataset design and labelling guidelines to training, fine-tuning, evaluation, deployment, monitoring, and retraining.
• Build and scale low-latency inference services for real‑time moderation using Python, PyTorch, Docker, Kubernetes, and streaming infrastructure such as Kafka.
• Define the metrics that actually matter for safety, including precision at a fixed recall, performance across languages and geographies, and the real cost of false positives on genuine users.
• Use LLMs and foundation models pragmatically for labelling, policy classifiers, and evaluation, while balancing cost, accuracy, and latency.
• Partner with Moderation and Policy teams to analyse misclassifications, close the data loop, and translate evolving safety policies into model behaviour.
• Continuously improve model performance, throughput, and serving costs through techniques such as distillation, optimization, and refactoring.
• Set engineering standards across the ML stack, reproducible training, experiment tracking, versioned datasets and models, and reliable production workflows.
• Mentor junior engineers and interns and help raise the overall engineering bar for the team.
• Stay on top of the latest research and tooling across Computer Vision, Audio, NLP, and Trust & Safety and turn the useful stuff into real‑world improvements.
• 4–6 years of experience as a Machine Learning Engineer, Applied Scientist, or similar role, with a strong track record of shipping ML models to production, not just research or notebooks.
• Python with deep hands‑on experience in PyTorch or TensorFlow.
• Strong depth in at least one of Computer Vision, Audio/Speech ML, or NLP, along with working familiarity across the others. This role touches all three.
• Experience with transformer architectures and modern pretrained backbones, including ViTs, wav2vec/Whisper‑style audio models, and BERT/LLM‑family text models, along with experience fine‑tuning them on domain‑specific data.
• Proven experience taking models into production — including serving, latency and throughput optimization, batching, GPU utilization, and monitoring live systems.
• Strong software engineering fundamentals including software design, testing, code reviews, and Git, along with practical experience using Docker and Kubernetes.
• Experience working with data at scale, including SQL, PostgreSQL, non‑relational databases, analytical databases such as ClickHouse, and streaming systems such as Kafka.
• Strong understanding of ML evaluation and experimentation — metric selection, class imbalance, drift detection, A/B testing, and human‑in‑the‑loop labelling pipelines.
• Good to have: Experience in Trust & Safety, content moderation, fraud/abuse detection, or Responsible AI.
• Good to have: Experience working with multilingual and code‑mixed data.
• Good to have: Experience with cloud ML infrastructure on AWS or GCP.
• Excellent problem‑solving skills and the ability to thrive in a fast‑paced, high‑ownership environment.
• Strong communication skills and the ability to collaborate effectively with cross‑functional teams.

A Quick Note on the Work

As part of building and improving our Trust & Safety systems, you may occasionally work with user‑reported content that can include sensitive material. We’re upfront about this during the interview process, and you’ll have the support and resources you need from the team.

Note - The 1st and 3rd Saturdays of every month are working days.

About FRND

FRND is redefining the way people connect by building social products that are engaging, safe, inclusive, and fun.

We’re a rapidly growing startup building for millions of users and solving unique challenges across social connection and entertainment. Our ambition is bold, and we're looking for people who want to build, experiment, and solve problems at scale.

Why FRND?

Impact at Scale: Work on products and initiatives that impact millions of users across India and international markets.

High Ownership: Take on meaningful problems and have the freedom to drive them from idea to execution.

Learn with the Best: Work closely with founders, leaders, and high‑performing teams while solving real business challenges.

Rewarding Journey: Competitive compensation, equity opportunities, and growth that matches your impact.

Work Hard, Have Fun: We're serious about building great products, but we also believe in enjoying the journey along the way.

Solve Interesting Problems: Work in an environment where curiosity, experimentation, and first‑principles thinking are encouraged.