IAS’ Lokesh Jain discusses his day-to-day work life as a senior MLOps engineer, and why it’s important to ‘stay curious’ in machine learning.
While Lokesh Jain started his career in software engineering and system design, a growing interest in data science and machine learning led him to venturing into machine learning operations (MLOps).
“I saw machine learning as a field where I could apply my engineering skills to new challenges like model deployment and data pipelines,” he says.
This interest led Jain to pursuing a master’s degree in machine learning, and today he holds the position of senior MLOps engineer at Integral Ad Science (IAS), where he specialises in designing large-scale data pipelines, ML platforms and deploying ML models into production.
“MLOps became the ideal blend, allowing me to leverage my technical foundation while managing the full life cycle of machine learning models, which is now central to my role at IAS.”
Tell me a bit about your role as a senior machine learning operations engineer.
As a senior machine learning operations engineer at IAS, I ensure the smooth deployment and scalability of machine learning models, while developing tools and frameworks for our machine learning platform. I design and implement robust data pipelines to support large-scale machine learning systems, allowing us to process vast amounts of multimedia data efficiently. I collaborate closely with data scientists, researchers and engineers to bring proof-of-concept models into production, ensuring their performance in real-time environments with high transaction volumes. By continuously improving the infrastructure, I help drive innovation and streamline operations, ensuring that our machine learning systems perform optimally at scale.
Image: Lokesh Jain
What skills do you use on a daily basis?
In my role as a senior machine learning operations engineer, I use a broad range of skills across software engineering, data engineering, DevOps, MLOps and cloud operations. Daily, I draw on my expertise in developing scalable systems, managing data pipelines and ensuring the smooth deployment of machine learning models. My work involves automating processes, developing tools, managing infrastructure for scalability and optimising the performance of models in production.
One area I didn’t expect to rely on as much before stepping into this role is communication skills. I collaborate with data scientists, engineers and researchers to bridge the gap between machine learning research and scalable production systems. Documenting processes and ensuring alignment across teams is crucial.
What are some of the biggest challenges you face when designing data pipelines that handle large-scale structured and unstructured data?
There are several things to think about when designing data pipelines for large-scale structured and unstructured data, especially for ad fraud detection. Ensuring scalability and performance is key, as ad fraud detection requires processing vast amounts of data. We apply distributed data processing techniques, which allows us to efficiently handle large datasets and scale our pipelines as needed. We also need to ensure we’re always maintaining data quality, factoring in the broad sources of datasets we analyse at scale continuously.
To address this, we use observability solutions to monitor pipeline performance, ensuring the system remains robust, scalable and responsive in detecting and preventing ad fraud.
‘Maintaining a healthy work-life balance is crucial, especially in a demanding field like AI’
How do you collaborate with data scientists to translate their requirements into operational solutions?
Collaboration with data scientists is a crucial part of my role. I work closely with them to understand the specific requirements for model development, such as the types of data needed, model architecture and performance metrics. My responsibility is to translate these requirements into scalable and efficient operational solutions. This collaboration bridges the gap between research and production, turning machine learning models into operational systems.
How do you maintain a healthy work-life balance?
Maintaining a healthy work-life balance is crucial, especially in a demanding field like AI. I set clear boundaries between work and personal time, ensuring I disconnect after work hours to recharge. Prioritising tasks and maintaining a structured routine help me stay focused, while scheduling short breaks throughout the day prevents burnout. Outside of work, I engage in activities that recharge me mentally and physically, like spending time with family, exercising or pursuing hobbies. I also practise mindfulness techniques, such as meditation or taking short walks, to manage stress and maintain focus. This balance of structure, downtime and self-care helps me stay productive and avoid burnout.
What do you enjoy most about the job?
What I enjoy most about my job as a senior MLOps engineer is the dynamic nature of the work and the opportunity to solve complex, evolving challenges. MLOps is a constantly changing field, with industries still defining its scope. This keeps the work exciting, as new tools, frameworks and methodologies regularly emerge to optimise machine learning workflows. The interdisciplinary nature of MLOps, blending data engineering, software engineering, machine learning and DevOps, provides a unique variety that I find deeply engaging. Each day brings fresh problems to solve, from building scalable pipelines to integrating cutting-edge technologies into production systems.
Another rewarding aspect is the significant impact my work has on driving business outcomes. By developing tools and frameworks that accelerate model training and ensure data quality, I help transform research into real-world, scalable solutions. Knowing that the systems I design are integral to the success of machine learning projects gives me a strong sense of fulfilment. The continuous learning and room for innovation in MLOps also allows me to explore a wide range of technologies, making the role both intellectually stimulating and professionally rewarding.
What advice would you give to early-career professionals trying to get into the machine learning space?
For those starting in tech or looking to break into the machine learning space, my first advice is to build a strong foundation in software engineering. Machine learning requires a solid grasp of algorithms, data structures and coding principles. Proficiency in languages like Python, along with skills in version control, testing and debugging, will equip you to tackle machine learning projects and navigate the more complex aspects of MLOps.
Secondly, stay curious and open to continuous learning. The machine learning landscape evolves rapidly, so it’s essential to keep up with the latest tools and frameworks. Given its interdisciplinary nature, learning about data engineering, statistics and DevOps is beneficial. Additionally, work on real-world projects – whether contributing to open-source, building your own models or collaborating on hackathons. This practical experience will accelerate your learning and give you something concrete to showcase. Ultimately, your ability to stay adaptable, explore new domains, and solve real-world challenges will be key to your success in this evolving field.
Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.
