Job Overview

American Express is hiring an Analyst - Data Science in Gurugram, Haryana, for an early-career role focused on modern artificial intelligence and machine learning systems.

The position is part of the data and analytics organization and offers hands-on exposure to model experimentation, inference, evaluation, deployment, and machine learning systems performance.

This role is suited to candidates who enjoy understanding how AI systems work beneath the surface and prefer practical engineering over notebook-only analysis. The work involves building prototypes, testing ideas, investigating failures, profiling systems, and turning promising research or experiments into maintainable implementations.

Key Responsibilities

• Experiment with large language models and multimodal models using Python and PyTorch.
• Develop prototypes as well as production-quality machine learning components.
• Optimize model inference performance and evaluate latency, throughput, batching, caching, GPU utilization, and memory consumption.
• Build evaluation frameworks to measure model quality and understand system behavior.
• Work with modern AI patterns such as embeddings, retrieval, reranking, structured generation, and agentic workflows.
• Deploy model-backed services and investigate their behavior under realistic workloads.
• Read relevant research papers and reproduce promising approaches where appropriate.
• Design controlled experiments and analyze failures across data, models, software, and infrastructure.
• Document technical findings, decisions, experiments, limitations, and lessons learned.

Required Skills

• Strong Python programming skills.
• Working knowledge of PyTorch.
• Understanding of modern neural-network architectures.
• Understanding of transformer concepts such as tokenization, embeddings, attention, sampling, and decoding.
• Ability to write maintainable production-quality software rather than relying only on notebooks.
• Familiarity with Linux, Git, Docker, APIs, and basic cloud infrastructure.
• Strong analytical, debugging, and problem-solving skills.
• Ability to design experiments, interpret measurements, and explain technical outcomes.

Preferred Skills

• Knowledge of machine learning systems and model serving.
• Familiarity with inference runtimes such as vLLM.
• Understanding of continuous batching and KV caching.
• Knowledge of quantization and mixed-precision computing.
• Understanding of GPU memory constraints and device placement.
• Experience with profiling and out-of-memory debugging.
• Knowledge of structured or constrained generation.
• Familiarity with CUDA, Triton, distributed systems, Kubernetes, or NCCL is an advantage.
• Research experience is helpful but not mandatory.

Education

A strong academic foundation in a quantitative discipline is preferred.

Relevant fields include:

• Computer Science
• Mathematics
• Statistics
• Engineering
• Physics
• Operations Research
• Other related quantitative disciplines

Experience

The position is designed for an early-career candidate and does not specify a numerical experience requirement.

Candidates can demonstrate suitability through substantial machine learning or systems projects, research or thesis work, implementation of research papers, open-source contributions, inference or training system development, technically serious side projects, or measured performance improvements.

The selection focus is on what candidates have built, how they approached technical decisions, what they measured, what failed, and what they learned.

Required Technologies

Python, PyTorch, Linux, Git, Docker, APIs, cloud infrastructure, large language models, multimodal models, neural networks, transformers, tokenization, embeddings, attention mechanisms, sampling, decoding, model serving, inference, vLLM, continuous batching, KV caching, quantization, GPU computing, mixed precision, device placement, profiling, structured generation, retrieval, reranking, and agentic systems.

Soft Skills

• Curiosity and willingness to explore unfamiliar technologies.
• Analytical thinking and structured problem solving.
• Strong technical communication.
• Persistence when investigating complex technical problems.
• Ability to test hypotheses and learn from failed experiments.
• Ability to document technical findings clearly.
• Ability to explain design decisions, performance measurements, and limitations.
• Ability to collaborate with technical stakeholders.

Benefits of Working in this Role

American Express lists career development and training opportunities along with flexible working arrangements based on role and business needs.

Eligible employees may receive competitive base salaries, bonus incentives, financial-well-being and retirement support, and location-dependent medical, dental, vision, life insurance, and disability benefits.

Other listed benefits include paid parental leave depending on location, access to wellness centers at eligible sites, confidential counseling through the Healthy Minds program, and opportunities to develop new skills and leadership capabilities.

Specific benefits depend on applicable policies and location.

Work Mode

Hybrid

Location: Gurugram, Haryana, India

Working arrangements can vary according to the role and business requirements.

Location

Gurugram, Haryana, India

Gurugram is a major technology and business hub with strong activity across software engineering, analytics, artificial intelligence, financial services, and enterprise technology.

Who Should Apply

This opportunity is suitable for early-career data science and machine learning professionals with strong technical foundations.

Candidates who have built substantial ML projects, implemented research papers, worked on inference or training systems, contributed to open source, or conducted meaningful performance experiments should consider applying.

Applicants should be comfortable with Python and PyTorch and should understand modern neural-network and transformer concepts.

Strong project evidence can be particularly valuable for candidates without extensive professional experience.

Career Growth

The role can provide a strong foundation in machine learning engineering, AI systems, model inference, evaluation, performance optimization, and production deployment.

Continued development can lead toward careers in machine learning engineering, AI engineering, ML systems, inference optimization, applied research, data science, or specialized AI platform engineering.

Application Advice

Build your resume around evidence rather than a long list of technologies.

Highlight substantial ML or systems projects and explain what you built, why you selected the approach, how you measured performance, and what you learned from failures.

Projects involving PyTorch, transformers, LLMs, inference optimization, retrieval, model serving, GPU performance, or distributed systems can be especially relevant.

Include research papers you reproduced, open-source contributions, benchmarks, experiments, or technical projects that demonstrate depth.

Be prepared to discuss design decisions, performance trade-offs, debugging methods, and the limitations of your work during the interview process.

Application Deadline

September 26, 2026

Apply now through SoftoJobs to explore this opportunity.