Within the Writer Highlight collection, TDS Editors chat with members of our group about their profession path in information science and AI, their writing, and their sources of inspiration. At the moment, we’re thrilled to share our dialog with Mariya Mansurova.
Mariya’s story is considered one of perpetual studying. Beginning with a robust basis in software program engineering, arithmetic, and physics, she’s spent extra thanover 12 years constructing experience in product analytics throughout industries, from search engines like google and yahoo and analytics platforms to fintech. Her distinctive path, together with hands-on expertise as a product supervisor, has given her a 360-degree view of how analytical groups may also help companies make the fitting choices.
Now serving as a Product Analytics Supervisor, she attracts vitality from discovering contemporary insights and revolutionary approaches. Every of her articles on In the direction of Knowledge Science displays her newest “aha!” second: a testomony to her perception that curiosity drives actual progress.
You’ve written extensively about agentic AI and frameworks like smolagents and LangGraph. What excites you most about this rising area?
I first began exploring generative AI largely out of curiosity and, admittedly, a little bit of FOMO. Everybody round me gave the impression to be utilizing LLMs or a minimum of speaking about them. So I carved out time to get hands-on, beginning with the very fundamentals like prompting strategies and LLM APIs. And the deeper I went, the extra excited I turned.
What fascinates me probably the most is how agentic techniques are shaping the way in which we dwell and work. I consider that this affect will solely proceed to develop over time. That’s why I take advantage of each probability to make use of agentic instruments like Copilot or Claude Desktop or construct my very own brokers utilizing applied sciences like smolagents, LangGraph or CrewAI.
Essentially the most impactful use case of Agentic AI for me has been coding. It’s genuinely spectacular how instruments like GitHub Copilot can enhance the velocity and the standard of your work. Whereas recent research from METR has questioned whether or not the effectivity positive aspects are really that substantial, I positively discover a distinction in my day-to-day work. It’s particularly useful with repetitive duties (like pivoting tables in SQL) or when working with unfamiliar applied sciences (like constructing an online app in TypeScript). General, I’d estimate a few 20% enhance in velocity. However this increase isn’t nearly productiveness; it’s a paradigm shift that additionally expands what feels potential. I consider that as agentic instruments proceed to evolve, we are going to see a rising effectivity hole between people and corporations which have realized methods to leverage these applied sciences and people who haven’t.
In the case of analytics, I’m particularly enthusiastic about computerized reporting brokers. Think about an AI that may pull the fitting information, create visualisations, carry out root trigger evaluation the place wanted, notice open questions and even create the primary draft of the presentation. That will be simply magical. I’ve constructed a prototype that generates such KPI narratives. And although there’s a major hole between the prototype and a manufacturing answer that works reliably, I consider we are going to get there.
You’ve written three articles underneath the “Practical Computer Simulations for Product Analysts” collection. What impressed that collection, and the way do you suppose simulation can reshape product analytics?
Simulation is a vastly underutilised software in product analytics. I wrote this collection to point out individuals how highly effective and accessible the simulations will be. In my day-to-day work, I hold encountering what-if questions like “What number of operational brokers will we’d like if we add this KYC management?” or “What’s the possible impression of launching this function in a brand new market?”. You may simulate any system, regardless of how advanced. So, simulations gave me a option to reply these questions quantitatively and pretty precisely, even when onerous information wasn’t but obtainable. So I’m hoping extra analysts will begin utilizing this method.
Simulations additionally shine when working with uncertainty and distributions. Personally, I want bootstrap strategies to memorising a protracted record of statistical formulation and significance standards. Simulating the method usually feels extra intuitive, and it’s much less error-prone in apply.
Lastly, I discover it fascinating how applied sciences have modified the way in which we do issues. With at the moment’s computing energy, the place any laptop computer can run 1000’s of simulations in minutes and even seconds, we are able to simply clear up issues that might have been difficult simply thirty years in the past. That’s a game-changer for analysts.
A number of of your posts give attention to transitioning LLM purposes from prototype to production. What frequent pitfalls do you see groups make throughout that section?
By means of apply, I’ve found there’s a major hole between LLM prototypes and manufacturing options that many groups underestimate. The commonest pitfall is treating prototypes as in the event that they’re already production-ready.
The prototype section will be deceptively clean. You may construct one thing purposeful in an hour or two, check it on a handful of examples, and really feel such as you’ve cracked the issue. Prototypes are nice instruments to show feasibility and get your workforce excited concerning the alternatives. However right here’s the place groups usually stumble: these early variations present no ensures round consistency, high quality, or security when going through numerous, real-world eventualities.
What I’ve realized is that profitable manufacturing deployment begins with rigorous analysis. Earlier than scaling something, you want clear definitions of what “good efficiency” seems to be like by way of accuracy, tone of voice, velocity and some other standards particular to your use case. Then it’s good to monitor these metrics repeatedly as you iterate, making certain you’re truly enhancing moderately than simply altering issues.
Consider it like software program testing: you wouldn’t ship code with out correct testing, and LLM purposes require the identical systematic method. This turns into particularly essential in regulated environments like fintech or healthcare, the place it’s good to reveal reliability not simply to your inside workforce however to compliance stakeholders as properly.
In these regulated areas, you’ll want complete monitoring, human-in-the-loop evaluation processes, and audit trails that may stand up to scrutiny. The infrastructure required to assist all of this usually takes way more growth time than constructing the unique MVP. That’s one thing that constantly surprises groups who focus totally on the core performance.
Your articles generally mix engineering rules with information science/analytics finest practices, corresponding to your “Top 10 engineering lessons every data analyst should know.” Do you suppose the road between information and engineering is blurring?
The position of an information analyst or an information scientist at the moment usually requires a mixture of expertise from a number of disciplines.
- We write code, so we share frequent floor with software program engineers.
- We assist product groups suppose by means of technique and make choices, so product administration expertise are helpful.
- We draw on statistics and information science to construct rigorous and complete analyses.
- And to make our narratives compelling and truly affect choices, we have to grasp the artwork of communication and visualisation.
Personally, I used to be fortunate to achieve numerous programming expertise early on, again at college and college. This background helped me tremendously in analytics: it elevated my effectivity, helped me collaborate higher with engineers and taught me methods to construct scalable and dependable options.
I strongly encourage analysts to undertake software program engineering finest practices. Issues like model management techniques, testing and code evaluation assist analytical groups to develop extra dependable processes and ship higher-quality outcomes. I don’t suppose the road between information and engineering is disappearing totally, however I do consider that analysts who embrace an engineering mindset can be far simpler in fashionable information groups.
You’ve explored each causal inference and cutting-edge LLM tuning strategies. Do you see these as a part of a shared toolkit or separate mindsets?
That’s truly an amazing query. I’m a robust believer that each one these instruments (from statistical strategies to fashionable ML strategies) belong in a single toolkit. As Robert Heinlein famously stated, “Specialisation is for bugs.”
I consider analysts as information wizards who assist their product groups clear up their issues utilizing no matter instruments match the very best: whether or not it’s constructing an LLM-powered classifier for NPS feedback, utilizing causal inference to make strategic choices, or constructing an online app to automate workflows.
Slightly than specialising in particular expertise, I want to give attention to the issue we’re fixing and hold the toolset as broad as potential. This mindset not solely results in higher outcomes but in addition fosters a steady studying tradition, which is important in at the moment’s fast-moving information trade.
You’ve coated a broad vary of subjects, from text embeddings and visualizations to simulation and multi AI agent. What writing behavior or guideline helps you retain your work so cohesive and approachable?
I normally write about subjects that excite me in the intervening time, both as a result of I’ve simply realized one thing new or had an fascinating dialogue with colleagues. My inspiration usually comes from on-line programs, books or my day-to-day duties.
After I write, I all the time take into consideration my viewers and the way this piece will be genuinely useful each for others and for my future self. I attempt to clarify all of the ideas clearly and depart breadcrumbs for anybody who desires to dig deeper. Over time, my weblog has turn into a private information base. I usually return to outdated posts: generally simply to repeat a code snippet, generally to share a useful resource with a colleague who’s engaged on one thing related.
As everyone knows, all the pieces in information is interconnected. Fixing a real-world downside usually requires a mixture of instruments and approaches. For instance, in case you’re estimating the impression of launching in a brand new market, you may use simulation for state of affairs evaluation, LLMs to discover buyer expectations, and visualisation to current the ultimate suggestion.
I attempt to replicate these connections in my writing. Applied sciences evolve by constructing on earlier breakthroughs, and understanding the foundations helps you go deeper. That’s why a lot of my posts reference one another, letting readers comply with their curiosity and uncover how completely different items match collectively.
Your articles are impressively structured, usually strolling readers from foundational ideas to superior implementations. What’s your course of for outlining a fancy piece earlier than you begin writing?
I consider I developed this fashion of presenting info in class, as these habits have deep roots. Because the e book The Tradition Map explains, completely different cultures fluctuate in how they construction communication. Some are concept-first (ranging from fundamentals and iteratively shifting to conclusions), whereas others are application-first (beginning with outcomes and diving deeper as wanted). I’ve positively internalised the concept-first method.
In apply, a lot of my articles are impressed by on-line programs. Whereas watching a course, I define the tough construction in parallel so I don’t neglect any vital nuances. I additionally notice down something that’s unclear and mark it for future studying or experimentation.
After the course, I begin interested by methods to apply this information to a sensible instance. I firmly consider you don’t really perceive one thing till you strive it your self. Though a lot of the programs have sensible examples, they’re usually too polished. So, solely whenever you apply the identical concepts in your personal use case will you run into edge circumstances and friction factors. For instance, the course may use OpenAI fashions, however I would wish to strive a neighborhood mannequin, or the default system immediate within the framework doesn’t work for my specific case and wishes tweaking.
As soon as I’ve a working instance, I transfer to writing. I want separate drafting from modifying. First, I give attention to getting all my concepts and code down with out worrying about grammar or tone. Then I shift into modifying mode: refining the construction, choosing the proper visuals, placing collectively the introduction, and highlighting the important thing takeaways.
Lastly, I learn the entire thing end-to-end from the start to catch something I’ve missed. Then I ask my companion to evaluation it. They usually deliver a contemporary perspective and level out issues I didn’t think about, which helps make the article extra complete and accessible.
To be taught extra about Mariya‘s work and keep up-to-date together with her newest articles, comply with her right here on TDS and on LinkedIn.
