In conversation with Sharadha, Director of Engineering at Yelp.

An entrepreneur, a contract engineer at various startups, and now the Director of Engineering, Data and Analytics Platform at Yelp, Sharadha Ramakrishnan has had one of the most incredible journeys through the world of engineering. A journey that we thought will inspire and prove resourceful for our community. So, we sat down with this brilliant technologist, super mom, and ardent advocate of women in engineering to learn more.

Here is an excerpt from our conversation:

Beginnings:

While it began as most traditional engineering careers do — science in high school and then software engineering after, her path grew a little untraditional after her masters. She founded her own startup after being inspired by the founder of the company she interned at during her final semester.

And thus, Monkey Creative Labs in Chennai came to be. “We were consulting with companies and attempting to build a social engine, building UDP protocols, microservices architecture and so on”. In 2013, she joined OrangeScape (now Kissflow) where she dabbled heavily in JavaScript. It was also when she got married and the couple decided to move back to the US.

“I decided that I should get some experience in the industry and then go back to being a founder if I liked”. And thus began her journey at Yelp.

“In 2017 I had the opportunity to build the metrics platform for Yelp. It was centralised metrics that eventually evolved into the data analytics platform. And, here I am today. We are a, engineering team of over 40 spread across geographies.

Woman in tech:

A disproportionate representation of women in tech and tech leadership is one of the industry’s biggest problems. Sharadha was no exception to the challenges that this put forth.

“Not being able to visualise myself in senior roles affected my confidence. The under-representation really deterred my progress and even led to me questioning my technical abilities.”

Even before she gave Yelp a shot, she went back to school for 9 months to learn Linux Kernel Architecture. “Yelp values diversity and inclusion above all else. The kind of opportunities and support that they offer under-represented minorities is phenomenal. Finding a workplace that is ready to offer you the platform and opportunities to showcase what you have potential for plays a big role in bringing women forward in the tech industry.” Sharadha has been consciously and relentlessly advocating for this.

On advice for women who want to be part of the industry:

When women in leadership positions are asked how they made it there, the answers we usually hear are ‘I was at the right time at the right place’. But Sharadha stands strongly against arbitrary answers like those.

“It is important to share your honest experiences so that other women can learn and make progress in their own journeys. One important piece of advice I’d give to women who are striving to make it in this domain is to find a sponsor. Not just a mentor, but a sponsor who can help find you a seat at the table. Someone who has complete faith in your work, will stand by you and your capabilities”.

But for that to happen, women need to make the leap and put themselves out there. “Write blogs, participate in discussions, join peer communities, network, meet people, talk about and discuss your work”, advises Sharadha.

One should also hold their company to higher standards for inclusivity. Women shouldn’t be afraid to speak up and stand their ground when it’s needed. They should also always support their peers and lift each other up. “Get involved in peer communities and employee resource groups. You will realise that the problems you are facing are systemic and you’re not alone. You will find the support you need and will also be able to extend that same support to someone else in need.”

Culture-first, always

Culture is not just about hiring for diversity and doling out perks. Consciously built culture supports their people no matter what.

“Culture is everything. It makes all the difference. And ensure you walk the talk. Then there will be no subcultures that prop up. Make sure employees feel heard, included and represented, then they will show up and deliver their best work.”

On sustaining that culture in the pandemic:

“The pandemic and remote work definitely made it difficult to reach out and stay connected with teams. But since we had already invested in building a culture-first workplace, the transition to the new normal was easier. When you have a strong culture that has been cultivated through inclusive and progressive experiences, it definitely stands the test of time and events like these.

“We have a philosophy for team meetings — it should be useful for everyone involved. Not just the management. So we set out some ground rules. There had to be a solid agenda when a meeting was set and our engineers were moderating them. So, there was always a clear idea of the purpose, takeaway and good involvement.”

Procedural vs. Agile

“We lean a lot towards agility. We’ve always gone the route of Percentage of Completion”. But a mix of both processes is the best way. Choosing which one depends on the need at hand. For their data team, they have adopted a more procedural method because a POC route might only deter what needs to be done. “Imagine we are thinking about a new log that’s going to analyse the product. Then it is critical that we start from data modelling, get it reviewed, move it to implementation within the app, then get it flowing through the data pipeline and so on.”

There is no objectively right or wrong process. It depends on evaluating what the use case and the need of the hour is and how the team is set up.

Reliability and availability engineering

“The importance of working on reactive things is something I speak about to engineers who want to join our data analytics team.” The engineers’ responsibility doesn’t end at writing and pushing code. They have to account for what happens after; what might happen if it breaks. going on call and fixing it.

It is fundamental and pivotal that every organization develops an incident response process. “By developing it, you contribute directly to engineering happiness. If there is no incident response in place, the engineer on call might not know how to respond and this will only lead to a string of mishaps from there on. A consciously developed incident response process should be non-negotiable. But of course, the foundational process like quality check and monitoring should be solid in the first place to even circumvent situations that might need incident response.”

The gaps between engineering and DevOps:

“At Yelp, we’ve been successfully mitigating these commonplace gaps because everyone is responsible for their own code.” Her engineers write, test and push their own code to production. The Site Reliability Engineering has been done away with. Every team is responsible for their own monitoring and alerts on their services.

Educating and enabling engineers to play both roles has helped her team tremendously. There is no one person shipping the code and another being responsible for production. The production engineering team can be depended on for support.

“We also practice a blameless culture. When there’s a mistake made, we learn from it. We do a postmortem on the issue, share the learning with the whole organization and progress together.”

On the Tech Stack for her division:

“We are a Python shop but some of the teams are polyglots. My team develops SDKs for data capture and we support Android, iOs, Java, Javascript, Python to name a few. Some engineers write Kotlin, Swift and Java alongside Python.”

Their data pipeline is primarily Kafka and there is an entire connector ecosystem around it. They use Redshift for analytics use cases and Cassandra and MySQL as data stores. They are primarily on AWS.

“We opened up our platform as PaaSTA — Platform as a Service Totally Awesome. It allows microservices deployment and provides everything that you need to set up with microservices. We’ve almost completely moved from the monolith.”

On technology evaluation and evolution:

An organization has to constantly keep evaluating and moving on to new technologies when there is a need. Decisions like this stem from various places. Two of the most important being:

• When there is a need to solve technical debt
• The need to be state-of-the-art

“Before the metrics platform, we worked as siloed teams. There was data being duplicated, data not being governed and it led to a lot of tribal knowledge. In 2016, we decided to build a single source of truth. We started by evaluating what was capable of supporting such a large purpose. We looked at what our peers are doing, assessed our needs thoroughly and brought in tech like Kafka and Flink to support it.”

“It is important to make the right ‘build vs. buy’ decision. If you are setting out to build a solution or tech, make sure that it impacts your process and business in the long run. If it doesn’t, the smarter thing would be to buy one that solves your problem.”

Building a data-driven culture

“Becoming a data-oriented or data-driven organization is more of a cultural shift than a technical one.” Having state-of-the-art technology and data doesn’t translate to the people within the organization using it. Becoming data-driven takes a significant mindset shift. “While healthy skepticism of data is good, learning to trust it is just as important.”

What can help propagate that mindset is a vision from the top; when the leadership is seen extensively relying on the data produced, it is bound to trickle down to the culture of the organization. Incentivizing people’s reliance on data with better user experience and keeping a lower barrier of entry also helps in encouraging a data-driven mindset.

The next big thing in data

“There’s always a lot happening with data. But I think the next phase is to figure out how to build an ecosystem of trust between data producers and consumers. Creating such an ecosystem will ensure that consumers are able to rely on the data that is being produced. On the data producer’s part it means data quality monitoring and establishing data governance, which are very up and coming in the domain.”

The conversation with Sharadha not only provided an insight into Yelp’s culture and technology but shed light on what is important in building a team as successful as hers.

We hope you found this useful. Head to our YT page to listen to the full conversation here: https://youtu.be/C4QP5xznRkw

Join CTOtalk on Linkedin for more such stories.